Skip to main content

Full text of "Fossil freshwater Emydid turtles of Florida"

See other formats


THE FOSSIL FRESHWATER EMYDID TURTLES 
OF FLORIDA 



By 

DALE ROBERT JACKSON 



A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF 
THE UNIVERSITY OF FLORIDA 
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE 
DEGREE OF DOCTOR OF PHILOSOPHY 



UNIVERSITY OF FLORIDA 
1977 



ACKNOWLEDGMENTS 

Walter Auffenberg introduced me to fossil turtles and David Webb 
increased my interests in paleontology. To both of these men, for their 
subsequent time and encouragement, I am deeply indebted. I am very 
grateful to the many people who reviewed parts of this paper or who shared 
their ideas and knowledge with me: Dennis Bramble, H. Kelly Brooks, 
Archie Carr, Stephen Christman, Richard Franz, Carter Gilbert, John 
Iverson, Howard Kochman, Carmine Lanciani, Frank Nordlie, Thomas Patton, 
Francis Rose, Roger Sanderson, Sylvia Scudder, Graig Shaak, Ernest 
Williams, and George Zug. 

For their generosity in providing fossils and comparative material 
I thank Walter Auffenberg, Walter Dalquest, James Dobie, Harold Dundee, 
J. Alan Holman, Farrish Jenkins, Curtis McKinney, Thomas Patton, Robert 
Purdy, John Waldrop, David Webb, and Stephen Windham. I am indebted to 
Nancy Halliday for instruction and assistance in preparing the illustrations 
and to Kay Purinton for the photographs. Lisa Megahee prepared the 
figures for Chapter IV and Kenneth Campbell the photographs for Figure 
17. Mike Frazier and Greg McDonald were especially conscientious in 
collecting and calling pertinent fossils to my attention, and I thank 
them for their efforts. 

To all others in the Florida State Museum and Department of Zoology 
who provided assistance and facilities, I extend my gratitude. 



TABLE OF CONTENTS 

Page 

ACKNOWLEDGMENTS ii 

TABLE OF CONTENTS iii 

ABSTRACT v 

INTRODUCTION 1 

CHAPTER 

I A PLEISTOCENE GMPTEMYS (REPTILIA: TESTUDINES) FROM 

THE SANTA FE RIVER OF FLORIDA k 

Description of Fossil Sites 4 

Description of Fossil Elements 5 

Discussion 8 

II EVOLUTION AND FOSSIL RECORD OF THE CHICKEN TURTLE 

DEIROCHELYS V/ITH A REEVALUATION OF THE GENUS 17 

Materials and Methods 18 

Fossil Localities 20 

Systematic Descriptions 20 

Discussion 37 

Relationships 39 

Distribution and Paleoecology k] 

III THE STATUS OF THE PLIOCENE TURTLES CHRYSEMYS CAELATA (HAY) 

AND CHRYSEMYS CARET ROSE AND WEAVER 68 

Taxonomic Considerations 69 

Relationships 72 

Discussion 7^ 

IV A REEXAMINATION OF THE CHRYSEMYS SCRIPTA GROUP BASED 

ON FOSSIL EVIDENCE 83 

Trachemus in the Pleistocene 85 

Tvachemys in the Upper PI iocene 88 

Systemat ics 88 

Paleoecology 96 

Systematic Conclusions 98 

Further Problems 101 



CHAPTER 

V PRE-PLIOCENE RECORDS OF THE GENUS CHEYSEMYS 105 

Materials and Methods 1 05 

Site Records for Chrysemys IO6 

1-75 : : : io6 

Seaboard Air Line Railroad Company, Switchyard B . . . 1 06 

Thomas Farm 106 

Discussion j 09 

APPENDIX- FOSSIL LOCALITIES CONTAI N I NG .Z)^//?^^^^!^ 112 

LITERATURE CITED 118 

BIOGRAPHICAL SKETCH 128 



Abstract of Dissertation Presented to the Graduate Council 

of the University of Florida in Partial Fulfillment of the Requirements 

for the Degree of Doctor of Philosophy 



THE FOSSIL FRESHWATER EMYDID TURTLES 
OF FLORIDA 

By 

Dale Robert Jackson 
August, 1977 



Chairman: Walter Auffenberg 
Major Department: Zoology 



Limited fossil evidence shows that a turtle of the genus Graptemys 
lived in peninsular Florida in the Santa Fe River during the Pleistocene. 
The small sample of available material, including elements of both 
Ranchol abrean and Blancan periods, indicates that this turtle occupied the 
river throughout most of the Pleistocene before becoming extinct in that 
drainage. The apparent affinities of this turtle with Recent G. 
barbouri of the Apalachicola River system bring up several geologic and 
zoogeograph ic considerations. 

Prior evidence of the fossil history of the monotypic genus 
Deirochelys is limited to a single Upper Pleistocene fragment and a number 
of Subrecent elements from Florida. On the basis of several morphological 
adaptations unusual among emydine turtles (e.g., neural bone width and rib 
structure), fossils from twenty Florida sites, ranging from Miocene to 
Subrecent in age, are referred to the genus Deirochelys . Evidence of the 
gradual evolution of a suite of characters associated with a gape-and-suck 
method of feeding is presented. The Middle Pliocene representative of the 



genus is recognized as a distinct species which, like the Miocene fossils, 
is intermediate between modern D. retioularia and less specialized 
emydines such as Chrysemys. 

Reexamination of type and referred materials indicates that C. 
carri Rose and Weaver is indistinguishable from Chrysemys oaelata (Hay). 
Chrysemys oaelata is a characteristic member of Hemphillian faunas of 
Florida and appears to be immediately ancestral to C. nelson-i. 

Fossil members of the Chrysemys soripta group (subgenus Trachemys) 
in Florida are reexamined and compared to fossil Chrysemys from other 
parts of North America. Designation of the Rancholabrean subspecies 
C. s. petrolei is confusing and biologically unrealistic. Cranial material 
of C. platymarginata affirms the Trachemys affinities of this Blancan 
turtle described from Florida. However, a morphological comparison of 
C. platymarginata with C. idahoensis , described from the Upper Pliocene 
of Idaho, reveals that the two may be conspecific. Additional material 
from the Great Plains supports this hypothesis. The phylogeny proposed 
in 1967 by Weaver and Rose, in which the C. soripta group is believed 
to be more closely related to the C. rvbriventris series than either is 
to the C. floridana series, is rejected in favor of that proposed by 
McDowell in 196^4, in which the last two series are more closely related 
to each other than either is to the C. soripta group. 

All pre-Pl iocene records of the genus Chrysemys are reexamined. 
Patton's 1969 report of Chrysemys from a nonmarine Oligocene deposit in 
Florida cannot be confirmed. Fragments assigned to the genus by Olsen 
from a Miocene deposit in the Florida panhandle actually represent a 
sea turtle and land tortoise. Chrysemys is present in the Thomas Farm 



Miocene fauna, although some material previously assigned to the genus 
may represent Deiroohelys. 



INTRODUCTION 

The family Emydidae, including nearly 30 living genera and 80 
species, is the largest of all turtle families. Its essent ial 1 y circum- 
global distribution (excluding Australia and subsaharan Africa) makes 
it the most widely distributed of all nonmarine turtle families. 
Emydids are especially well-represented in southeastern Asia and the 
southeastern United States and are the dominant family of turtles in 
North America (Auffenberg, 197^; Mount, 1975). 

A major step in understanding evolution of the modern chelonian 
fauna lies with clarification of the relationships and evolutionary 
processes that have occurred within the Emydidae. Although in the last 
two decades considerable significant comparative work in this field has 
been accomplished through the techniques of serology {e.g., McKown, 
1972; Merkle, 1975), cytogenetics {e.g., McKown, 1972; Bickham, 1975; 
Bickham and Baker, 1976), and comparative morphology {e.g., McDowell, 
196A; Waagen, 1972; Bramble, 197^*), inadequate attention has been paid 
to the direct evidence provided by the fossil record. 

Freshwater emydid turtles in the southeastern United States, and 
especially those in Florida, are wel 1 -represented in vertebrate fossil 
deposits. This paper attempts to reevaluate, expand, and coordinate 
our knowledge of these turtles as reflected by their fossil record. 
In addition to systematic and evolutionary cons iderat ions, ecolog ic 
and biogeographic implications are discussed. The geographic domain 
of this paper is Florida, although the relationships of turtles from 



this area cannot be meaningfully discussed without comparisons to 
turtles from other regions of North and Central America. 

Although probably originating in the Late Mesozoic, the Emydidae 
are unknown as fossils before the Tertiary. The family Is well-repre- 
sented in several Eocene formations in Western North America, primarily 
by the presumably extinct genus Eohmatemys. Most modern genera remain 
unknown as fossils until the Miocene or later. The geologic youth of 
Florida restricts the present work to Oligocene and younger strata; 
hence, early evolutionary paths within the family will not be found 
here. 

By far the most important early thrust in chelonian paleontology 
was Oliver Perry Hay's (1908) "Fossil turtles of North America," in 
which he discussed extensive Pleistocene and older material from 
Florida. Since then, little work was done in Florida until the late 
1950*5 and 1960's, when there occurred a resurgence of interest in 
fossil emydids. Most significant are the works of Weaver and Rose 
(1967) and Weaver and Robertson (1967), which deal with the genus 
Chrysemys, and a series of papers by Mil stead and Auffenberg (Milstead, 
1956, 1967, 1969; Auffenberg, 1958; Milstead and Tinkle, I967) on box 
turtles (Terrapene) . As the latter have been thoroughly studied, I 
shall not deal further with Terrapene. 

The following work consists of five sections each dealing with 
a distinct taxonomic unit. The first section, details the first known 
occurrence of Gvaptemys harbouri in the fossil record and the first 
fossil record of the genus in Florida. The second section describes 
a remarkable collection of fossils that clearly outline the direction 
of evolution in the previously poorly known genus, Deirochelys. The 



third revises the relationships of two Middle Pliocene turtles of the 
genus Chnjsemys. The fourth presents a new interpretation of the 
Chryserrys scripta complex based on a comparison of fossils from Florida 
with those from the midcont inent , and the fifth reexamines the Miocene 
and Oligocene fossils previously assigned to the genus Chryserrys. 

As elements of the shell are by far the most frequently found and 
readily recognizable turtle fossils, it is primarily with these that I 
have worked. Many adaptive characters are reflected by shell morphology. 
Cranial material is used when available, although it is unfortunately 
scarce in most fossil deposits. Chelonian limb elements are generally 
of little systematic value at the specific level and are very rarely 
found in association with shells in Florida deposits. 



CHAPTER I 
A PLEISTOCENE GRAPTEMYS (REPTILIA: TESTUDINES) 
FROM THE SANTA FE RIVER OF FLORIDA 



Prior to this study the genus Graptemys has not been reported in 
the southeastern United States east of the Apalachicola River system 
in western Georgia and the Florida panhandle. All valid records indi- 
cate that the eastern-most species, G. barbouri , is presently endemic 
to the Apalachicola drainage (Cagle, 1952; Carr, 1952; Dobie, 1972; 
Wharton et al . , 1973). Fossil elements from the Santa Fe River in 
northern peninsular Florida now prove that a turtle of this genus did 
occur there during the PI io-Pl ei stocene. The close alliance of this 
turtle with modern G. barbouri suggests the importance of Pleistocene 
physiographic changes to the distribution of these animals. The 
discovery of this fossil form adds to our limited knowledge of the 
fossil history of the genus, previously summarized by McKown (1972). 

Description of Fossil Sites 

The Santa Fe River in northern peninsular Florida has previously 
been described by Hellier (1967); it is presently one of the major 
tributaries of the Suwannee River. All fossils were obtained from two 
adjacent sites in the Santa Fe at the Columbia/Gilchrist county line 
(29°50'N, 82°42'W), well downstream from the subterranean portion of 
the stream's course. Both sites are bottom deposits with reworked 
bone. Santa Fe I is a heterochron i c deposit containing material from 
both the Rancholabrean (late Pleistocene) and Blancan (Upper Pliocene) 



periods; it normally lies under 6 to 8 m of water. Fossils representing 
the two ages from this site may generally be distinguished by their 
state of preservation; Blancan elements are often a glossy black in 
contrast to the coarse brown Rancholabrean material. Santa Fe II , which 
lies in 2.5 to 3 m of water approximately 100 m downstream from Santa 
Fe I, contains only Rancholabrean material. Site ages have been well- 
established on the basis of their mammalian faunas (Hibbard ,et .a!., 
1965; Webb, 197^). 

Description of Fossil Elements 

Three elements positively representing the genus Graptemys are 
available: a third neural, a nearly-complete nuchal bone, and the major 
portion of a mandible (Fig. 1). A right hyoplastron representing either 
a Graptemys or a small Chrysemys concinna is also described. All 
specimens are in the Florida State Museum (UF) and the Timberlane 
Research Organization (TRO). 

Nuchal bone . A nearly-complete nuchal bone (UF 10572) from Santa 
Fe I was collected by B. Waller and R. Allen in 1963. The very broad, 
short nuchal scute is characteristic of Graptemys. Although the dorsal 
midline of the bone is conspicuously elevated, the distinct keel usually 
present in G. barbouri is lacking. However, Roger Sanderson (pers. 
comm. ) , after examining several hundred G. barbouri , believes this 
character to be of little taxonomic significance. Dimensions of the 
fossil are width of anterior border--approx. 35-6 mm; max width-- 
approx. 69.8 mm; length of anterolateral border--8.1 mm; max length-- 
47.8 mm; length of nuchal scute — 3.6 mm. Estimates based on these 
measurements place the fossil turtle between 210 mm and 230 mm cara- 



pace length (CL), corresponding to a plastron length (PL) of 180 mm 

to 197 mm; carapace width (CW) is estimated at roughly 190 mm to 200 

mm. These dimensions correspond to those of Recent adult female 

G. barbouri although the carapacial width is relatively greater. 

State of preservation indicates that this turtle is probably Blancan in 

age. 

Neural bone . A large Graptemys neural bone (TRO 100) was found 
in 1961 at Santa Fe 11 by J. Waldrop and D. Bell. The distinct keel 
ending abruptly in a blunt knob midway back along the midline of the 
neural indicates that this is the third neural of a G. barbouri-] \ke 
turtle. Dimensions of the bone are max length--30.6 mm, max width 
(anterior margin, lateral border) --30. 5 mm; min width (posterior margin, 
lateral border)-- 18. mm; max thickness, lateral border--10.0 mm. 
Except for the more rapid tapering from anterior to posterior borders, 
the neural appears to be that of a very large G. barbouri (estimated 
CL = 270 mm to 280 mm; PL = 230 mm to 240 mm). 

Mandible . The only available data for the partial lower jaw 
(US 19161) is the collection site, Santa Fe II. The broad alveolar 
surfaces as well as the size of the jaw indicate that this turtle had 
an extremely large head modified for crushing hard food such as mollusks. 
This structure and habit is characteristic of females of the sexually- 
dimorphic extant forms, G. barbouri and G. pulohra. The fossil mandible 
appears almost identical to that of a Recent adult female G. barbouri, 
except that the alveolar surfaces of the fossil are less expanded 
posteriorly (possibly a function of ontogenetic change), and the two 
halves of the jaw meet at a slightly wider angle than in the Recent 



form. Dimensions of the fossil elements are approx. max width 
(excluding articulating surface) --5'^ mm; max width of alveolar surface-- 
18.6 mm. Comparison with Recent G. hai'bour'-i yields an estimated 
carapace length for this individual of 230 mm to 290 mm (assuming 
skull rshell ratios to be approximately the same in Pleistocene and 
Recent forms) . 

Hyp opl a stron . The only data for the right hypoplastron (UF 192^46) 
is Santa Fe 1. The thin, flat nature of the piece, the very narrow zone 
of scute overlap on the dorsal surface, and the straight anterior 
border {i.e., the absence of an obvious concavity into which the ento- 
plastron fits, as in most Chrysemys) Indicate that this element may be 
from Grapteniys . However, l:ypoplastra of some individuals of Chrysemys 
eoYicinria, a common turtle of both the Pleistocene and present in the 
Santa Fe River, also fit this description, and the element may belong 
to a small Individual of that species. The dimensions of the bone are: 
max length--62.7 mm; width from midline to anterior, ventral edge of 
axillary notch--A8.5 mm; max thick^-iess along hyo-hypopl astral suture-- 
k.k mm. State of preservation Indicates a probable Blancan age. 

Based on available material it is concluded that the Santa Fe 
Grcwtemys vjas a sexually-dimorphic form (or forms) very similar to G. 
harhouvi in both structure and habits. Maximum adult body size was 
probably slightly greater than that attained by modern G. barbouri. 
The shell may have been somewhat broader and the keel less pronounced 
than in extant G. barbouri . Pending acquisition of further material 
which inay show the Blancan, Ranchol abrean , and Recent forms to represent 
two or even three distinct forms, all the Santa Fe Graptetnys are 



tentatively referred to Graptemys cf. G. harbouri Carr and Marchand. 
Erecting further names for these allochronic forms is presently un- 
warranted. 



Di scussion 



Accounting for the presence of G. harbouri in the Suwanee drainage 
during the Pleistocene requires consideration of several geologic and 
zoogeographic phenomena. Mainly because Graptemys rarely if ever travels 
on land (McKown, 1972), endemism to a single river system is character- 
istic of several species of the genus--G. versa, G. ooulifera, G. 
flavimaculata, and for all intents and purposes, G. nigrinoda (Cagle, 
195/4; Folkerts and Mount, 1969; Conant, 1975). Since Dobie (1972) 
eliminated Cagle's (1952) apparently-erroneous record for the Escambia 
River in western Florida, all remaining records indicate that G. 
barbouri likewise developed as an endemic species in the Apalachicola 
River system. The presence of a Pleistocene turtle, here believed to 
be conspecific with Recent G. barhouri (although the problem remains 
essentially the same even if it were later shown to be a distinct 
species of common ancestry), in a drainage system which empties into the 
Gulf of Mexico approximately I85 km from the mouth of the Apalachicola 
requires comment. Several explanations are suggested. 

The eustatic changes in Pi io-Pl ei stocene sea levels, corresponding 
to the formation and melting of the Ice Age glaciers, are well-known 
phenomena. During glacial periods the retreating sea exposed a much 
larger expanse of the Floridian Plateau (that portion of the continental 
shelf surrounding Florida and extending far out into the Gulf-Cooke, 
191,5) than is now above sea level. Frey (1965) has stated that this 



eustatic lowering of the sea level reached a maximum of 120 m below 
present sea level, and during the Wisconsinan glacial episode some 
20,000-18,000 years ago exposed up to a 210 km wide stretch of con- 
tinental shelf (Fig. 2). Donn et al. (1962) have calculated that 
minimum sea level came even earlier, in the lllinoian glacial period, 
when the sea dropped to between ]hO m and 1 60 m below its present 
level. Conceivably such a drop in sea level could bring about the 
confluence of many rivers which now empty directly into the Gulf. Con- 
tinuation of the Apalachicola and Suwannee Rivers beyond their present 
mouths would create a hypothetical junction between them (Swift, 1970) 
near the expected western-most coastline of Pleistocene central 
peninsular Florida. However, the absence of large numbers of fish 
species common to both rivers as well as intervening drainages discredits 
this idea. Nevertheless, even if convergence of the two rivers did not 
occur, a lowland marsh between the two could have allowed turtles to 
migrate from one stream to the other. If this were the case the 
occurrence of a G. barbouvi- \\ ke. turtle in two presently widely sep- 
arated rivers could thus be explained. Furthermore, if the present 
absence of Graptemys fossils from the Suwannee north of its junction 
with the Santa Fe reflects a real situation, then it may be that the 
ancestral (upper) Suwannee was captured by a stream (the present lower 
Suwannee, including the Santa Fe) eroding eastward from the Gulf of 
Mexico during the PI io-Pl ei stocene (Vernon, 1951; Brooks, 1966). 
Whether this stream was confluent at one time with the Apalachicola is 
not presently known. Evidence shows that prior to this time the 
ancestral Suwannee likely emptied into the Gulf south of its present 



10 



mouth via connection with the present-day Waccasassa River (Vernon, 
1951 ; White, 1970). 

A second possible migratory route which may have been followed by 
the ancestral G. barbouri involves the upper reaches of these rivers. 
Although Graptemys rarely wanders overland it is conceivable that under 
some conditions (e.g., flooding) a few individuals may move to an 
adjacent stream. Figure 2 reveals the proximity of the Flint River, a 
major tributary of the Apalachicola in which G. barbouri is known to 
occur (Cagle, 1952; Wharton et al., 1973; R. Franz, pers. comm.), to the 
upper reaches of the Wi thl acoochee and Alapaha Rivers, both of which 
empty into the Suwannee. The distances may have been even less during 
the Pleistocene had the waters been significantly higher or the courses 
of the rivers different than they are now; the latter possibility is 
especially likely in this area of sand- 1 imestone substrate. Further- 
more, certain fishes common to the two drainages suggest direct 
communication at some time in the past between the upper reaches of 
these two rivers (C. Gilbert, pers. comm.). Yerger and Relyea (I968) 
explain the distribution of laixzlurus serracanthus --knovjn only from 
the Apalachicola, Ochlockonee, and Suwannee drainages — by stream piracy 
following closure of the Suwannee Straits (Pl io-Plei stocene) . They 
suggest that stream capture between tributaries of the Ochlockonee 
(until relatively recently a part of the Apalachicola drainage) and 
the upper Suwannee may have accounted for a westward migration of this 
fish from the Suwannee into the Apalachicola. It is likely that stream 
capture between tributaries of these same rivers or between the Flint 
and upper Suwannee may account for the occurrence of G. barbouri in 
these two drainages. Its apparent absence (both Pleistocene and Recent) 



11 



from the upper Suwannee and Ochlockonee may reflect inadequate collect- 
ing or unsuitable habitat. 

A third possibility, not necessarily exclusive of either of the 
preceding two, is that G. harbouri has not always been restricted to 
one or two widely separated rivers but was more widespread than pre- 
viously thought. The long period of relatively stable environmental 
conditions during the Blancan may have been conducive to the expansion 
of G. barbouri' s range throughout much of the seemingly favorable 
limestone-underlain habitat between the two rivers. Thus it is postu- 
lated that at various times G. barbouri occurred (though perhaps in 
relatively low densities) in most or all rivers between the Apalachicola 
and Suwannee. In addition to explaining the presence of this turtle in 
two widely separated rivers as simply its occurrence at the ends of a 
nearly-continuous range, the idea also seems compatible with the dis- 
tribution of the remaining southeastern members of the genus. The 
occurrence of G. kohni in the rivers of eastern Texas and Louisiana 
(Conant, 1975), G. pulchra from the Pearl River to the Escambia and 
Yellow Rivers in western Florida (Dobie, 1972), and G. barbouri from 
the Apalachicola to the Suwannee River, would provide a nearly- 
continuous range of large, ; --ixual ly-d imorphic Graptemys of the wide- 
headed female line (Cagle, 1953; McKown, 1972) in the river systems 
draining the entire northern shore of the Gulf of Mexico. The only 
apparent gap is the Choctawhatchee River System between the ranges of 
G. pulchra and G. barbouri (Dobie, 1972) (Fig. 2). 

Ultimately the solution to this question may lie with the collec- 
tion of further fossil evidence and with detailed studies of the dis- 



12 



tributions of other lotic organisms. Current investigations on fish 
and snail faunas of these and other river systems in the southeastern 
United States may help provide an answer. 

The reasons for extinction of the Santa Fe Graptemys remain 
speculative. At present the river seems sufficiently similar to Carr 
and Marchand's (19^2) description of ideal G. barbouri habitat to support 
a population of these turtles; this seems to be further indicated by 
the fact that lotalurus serraaanthus , which also seems to prefer rocky, 
clear streams with moderate flow, still inhabits the Santa Fe today 
(Yerger and Relyea, I968). From the paucity of fossil material Graptemys 
was apparently never common in the Santa Fe. In comparison to the few 
Graptemys fossils over 500 elements of Chrysemys aonainna, C. nelsoni, 
and C. scripta have been collected from the same sites. Even though 
C. oonoinna is abundant in the Santa Fe today its habits are sufficiently 
distinct (and presumably were in the Pleistocene) from those of G. 
barbouri that significant competition between the two turtles would 
have been unlikely. More probably, climatic changes or physical changes 
in the topography of the land or in the river itself made the habitat 
unsuitable at some time in the past for the small population of Graptemys, 
or possibly for the mollusks upon which the females presumably fed. 



Figure 1. Fossil Graptemys elements from the Santa Fe River, 

Florida. (A) nuchal; (B) mandible; (C) , (D) third neural 
dorsal and lateral aspects. 



14 







Figure 2. Southeastern United States and Gulf of Mexico, showing 
range of Graptemys pulahra and G. barbouri . Diagonal 
hatching, approximate range of G. pulahra (northern 
boundary indefinite); horizontal hatching, reported range 
of G. bavbouvi; triangle, Santa Fe River sites I and II; 
dotted line, 120 - meter contour showing approximate 
coastline during maximum eustatic lowering of sea level 
in the Quaternary (after Frey, I965); rivers mentioned 
in text: 

1- Pearl 6. Apalachicola 10. Alapaha 

2. Escambia 7. Ochlockonee 11. Suwannee 

3- Yellow 8. Fl int 12. Santa Fe 

h. Choctawhatchee 9. Wi thlacoochee I3. Waccasassa 

5. Chipola 



]6 



n 




CHAPTER I I 
EVOLUTION AND FOSSIL RECORD OF THE CHICKEN TURTLE 
DEIEOCHELYS WITH A REEVALUATION OF THE GENUS 



The evolutionary history of the monotypic genus Deiroahelys is 
one of the more enigmatic chapters in our knowledge of North American 
emydine turtles. Previous workers (Carr, 1952; Loveridge and Williams, 
1957; McDowell, 196^) have generally agreed that Deiroahelys is a highly 
specialized derivative of the genus Chrysemys (sensu McDowell, 196^*). 
Furthermore, Baur's (I889) suggestion of a close phylogenetic relation- 
ship between Deiroahelys and another North American monotypic emydine 
genus, Emydoidea, has been supported by most subsequent workers 
(Loveridge and Williams, 1957; Jackson, 1959; McDowell, 1964; Zug and 
Schwartz, 1971). Recently Waagen (1972) and Bramble (1974) have cast 
doubt on this idea based on their respective studies of musk glands and 
shel 1 mechan ics . 

The fossil record has been of no help in these matters to date. 
Prior knowledge of the fossil history of the genus Deiroahelys is 
limited to description of one partial nuchal bone from the Upper 
Pleistocene of Florida (Jackson, 1964, 1974a) and to mention of the 
presence of D. reticularia in a Subrecent Florida site (Hirschfeld, 
1968). Jackson (1974a) suggested that the Pleistocene element 
representsa turtle which is conspecific with modern D. reticularia. 
All other fossils assigned to the genus, i.e. Deiroahelys floridana 
Hay and Trachemys jarmani Hay (Hay, I908; Weaver and Robertson, I967), 
actually represent the genus Chrysemys (Jackson, 1964, 1974a). 

17 



This paper examines material referable to the genus Deirochelys 
from one Miocene, five Pliocene, twelve PI ei stocene, and two Sub recent sites, 
all in Florida. The Miocene fossils are the oldest known representatives 
of the genus. Two species of Deirochelys, one new, are recognized as 
fossils. As will be shown the major course of evolution within Deiro- 
chelys has been the extreme elongation of the head and neck, a condition 
achieved by only one other emydine genus [Emydoidea) and presumably 
developed as a trophic specialization. The accompanying cervical 
musculature hypertrophy has necessitated further structural modifications 
of the shell and vertebral column. It is for this reason that in trac- 
ing the evolution of the genus I dwell primarily upon this cervico- 
cranial elongation and associated morphological modifications (e.g., 
changes in neural bone width and rib and vertebral structures), to 
which I collectively refer hereafter as a single "character suite." 

Materials and Methods 

All fossil specimens except those from Waccasassa River and a few 
from Thomas Farm are part of the vertebrate paleontology collection of 
the Florida State Museum (UF) ; the Waccasassa River I specimens are 
from the Timberlane Research Organization (TRO) , Lake Wales, Florida, 
while some of the Thomas Farm fossils are from the collections of the 
Museum of Comparative Zoology, Harvard University (MCZ) . Comparative 
skeletal material was examined from the herpetology collection of the 
Florida State Museum (UF), the National Museum of Natural History 
(USNM), and my personal collection (DRJ). Extant specimens examined 
were Deirochelys reticularia : DRJ 264, 266, 270, 27^4, 278-280, UF 



19 



li»20, llkk, \hlkk, USNM 11610, 11615, 29^77, 29584, 62219, 80965, 
95789; Emydoidea hlandingii: UF 1^42^*9, 18931; Chelydra serpentina: 
DRJ 253; Chelys fimbriata: UF 21977- 

A shell thickness index (STI) was determined for most fossils. 
Thicknesses of fossil shell elements were measured and divided by 
corresponding measurements of a series (N = 10) of Recent adult D. 
retioularia of corresponding size, or by linear extrapolations to pro- 
duce such if no modern turtle of sufficient size could be found. As 
the relationship between shell thickness and body length may be 
asymptotic rather than strictly linear, the STI values given for the 
largest fossils may actually be underestimates. There was little 
individual STI variation among the Recent turtles when corrected for 
differences in body length. More medial elements (neural bones and 
proximal ends of pleural bones) generally yielded slightly higher STI 
values than peripheral elements (peripheral, pygal , and nuchal bones), 
indicating that increase in shell thickness is not necessarily pro- 
portional for all parts of the same shell. Medial edges of peripheral 
elements were measured to reduce this discrepancy. 

An index of free rib length (width of rib canal) was determined 
by dividing the stra ight- 1 ine distance from the proximal tip of the 
pleural bone to its union with the rib by the width of the pleural bone 
at the level of the union. The fragmented condition of most of the 
fossils necessitated the use of pleural bone width rather than length. 
In comparing neural and pleural bones of fossil Deiroohelys with 
those of Recent turtles, it is necessary to determine which of the 
eight neural or pleural bones the fossils represent. The presence 



20 



and position of scute sulci as well as the relative proportions of the 
anterolateral and posterolateral borders of the bones usually make this 
possible. Because of the relatively great width and frequent anomalies 
of the posterior neural bones of most emydine turtles, these bones are 
of little taxonomic value. 

All measurements are maximum and given in mm. 

Fossi 1 Loca 1 i t ies 

The appendix provides an annotated list of Florida localities that 
have yielded fossil Deivochelys mentioned in this paper. Reference is 
made to other publications in which stratigraphy, pa 1 eoecology , and 
correlative age of each of these deposits is described in detail. 
Figure 3 shows the geographic distribution of these sites. 

Systematic Descriptions 

All past descriptions of the genus Beivochelys (Agassiz, 1857; 
Baur, 1889; White, 1929; Schwartz, I956; Jackson, 1959; McDowell, 196^1; 
Zug and Schwartz, 1971) have necessarily been drawn solely from the 
single extant species, D. reticularia. Hence, many characters which ' 
would have been more appropriately designated as specific characters, 
particularly those involving color pattern, have been incorporated 
into the definition of the genus. Therefore, in order to accommodate 
the fossil members of the genus it is necessary to relegate many of 
the generic characters, including all references to color pattern, to 
specific level. Additionally, an examination of osteological char- 
acters through time reveals phylogenetic changes within the genus that 



21 



may be used to distinguish certain allochronic forms. For these reasons 
I find it necessary to give a brief systematic reevaluation of the genus 
as a prelude to a formal description of the fossil forms. The present 
chronologically-expanded definition of the genus, like those of Baur 
(1889), White (1929), Jackson (1959) and McDowell (1964), is based 
solely on osteological characters. As fossil skull material is unknown, 
all skull characters are drawn from modern Z). retioularia. Schwartz 
(1957) gives a brief but adequate account of the taxonomic history of 
the genus. 

Fami 1y Emyd idae 

Subfamily Emydinae 

Genus Deiroahetys Agassfz 

To the generic synonymy given by Zug and Schwartz (1971) should 
be added the following entry: 

Hiroohelys Beyer, 1900: k5- 

Type . Testudo reticularia Latreille. 

Referred species . Beivochely s ret-iculax'ia , the only extant 
species, presently distributed throughout the southeastern United 
States and known from the Pleistocene of Florida; Deiroohelys carvi , 
n. sp.. Pliocene Alachua clays of Florida, Hemphill ian age. 

Def ini t ion . Shell elongate to subovate in adults; carapace 
elliptical or cuneiform in outline and usually sculptured with fine 
parallel ridges or scales (Fig. k) ; anterior edge of nuchal bone 
generally truncate and acuminate; lateral sulci of nuchal scute 
usually parallel above and below; nuchal scute usually two to three 



22 



times longer than wide above, approximately as wide as long below; 
nuchal bone overlapped only by small corner of first costal scute or 
not at all; vertebral scutes as wide as long; neural bones hexagonal, 
short-sided in front; first neural bone circular to subovate in out- 
line; other neural bones generally as wide or wider than long (Fig. k) ; 
peripheral bones unnotched; pygal bone approximately paral 1 el -s ided 
with a shallow mesial notch; ribs dorsally free from pleural bones 
well below proximal ends of pleurals, their free portions slender and 
bowed ventral ly (Fig. 5) accommodating the enlarged trunk vertebral 
muscle complex (Shah, 1963). 

Plastron usually considerably narrower than carapace, akinetic, 
and firmly united to carapace by a high bony bridge and plastral 
buttresses; inguinal scutes large [contrary to Holman's (1967) state- 
ment that they are absent]; plastron smooth ventral ly or with traces 
of sculpturing similar to but less pronounced than that of carapace; 
entoplastron usually anterior to humeropectoral sulcus and overlapped 
by gular scutes for approximately one third of length. 

Skull and second through seventh cervical vertebrae elongate; 
neural spines of anterior thoracic vertebrae laterally compressed as 
vertical sheets (Fig. 6); triturating surfaces of maxilla and mandible 
narrow, without ridges; beak never hooked; interorbital width very 
narrow, less than one-half diameter of orbit; palate decidedly flat; 
posterior palatine foramina much larger than foramina orbi to-nasal e 
(Gaffney, 1972) (= anterior palatine foramina of Hoffman, I89O); 
temporal arcade complete; quadrate nearly enclosing stapes; hyoid 
apparatus strongly developed, lateral horn length at least as great as 
skull width; cervical musculature as described by Shah (1963). 



23 



A specialization of the genus almost certainly related to the 
elongate neck and hypertrophied vertebral musculature is the modification 
of the spinal column. The differences between DeivochBlys and more 
primitive emydines {Chrysemys, Echmatemys) , summarized in Table 1 and 
Figure 6, are most conspicuous in the first four thoracic vertebrae. 
In both forms ribs attach intercentral 1 y and the thoracic vertebrae 
are united by their neural spines to the overlying neural bones. The 
net effect of these modifications in Deiroahelys has been to move the 
rib attachment ventral ly (away from the carapace), allowing for the 
hypertrophied trunk vertebral musculature without changing the distance 
of the spinal cord from the ventral surface of the carapace. 

Deiroehelys reticularia (Latreille) 
Chicken Turtle 

The only addition to the species synonymy listed by Zug and 
Schwartz (1971) is: 

Hirochelys reticulata Beyer, 1900:^45. 

Type : The type was formerly in the collection of the French 
Museum National d'Histoire Naturelle but is now considered lost 
(Schwartz, 1956). Schwartz (1956) described a neotype^ and neoallo- 
type from the vicinity of the original type locality. 

Type local i ty . Restricted by Harper (19^0) to the vicinity of 
Charleston, South Carolina. 

Diagnos i s : A Dei-vochelys characterized by relatively low length: 
width ratios for third through fifth neural bones (means, 0.6 to 0.7; 



2k 



Table 1. Comparison of the thoracic vertebrae of 

Deiroahelys and Chrysemys . 



Character 



Chrysemys 



Neural spines low and robust 



Deirochetys 



laterally compressed as 
vertical sheets 



Centra 



narrowest ventral ly; 
not compressed 



narrowest dorsal ly; 
dorsoventral ly compressed 
ventral surfaceswide and 
f 1 attened 



S i te of rib 
attachment 
to vertebra 



expanded dorsal 
region of 
centra 



expanded ventral region 
of centra 



25 



Fig. 7) and relatively great length of free portions of dorsal ribs 
(Fig. 8); coloration as described by Schwartz (1956) with notation 
that the yellow forelimb band is usually but not always wide; neck 
nearly as long as plastron; usual pattern of cervical central 
articulation (perhaps a generic character): (2( (3( (4) )5) )6) )7( 
(8) (Williams, 1950; Jackson, I97^b). 

Description of fossil material . The following fossils, listed in 
reversed chronologic order by site, are here assigned to D. ret-Loularia, 

Nichoi's Hammock: contains more D. ret-icularia than any other post- 
Pliocene site; 75 carapacial elements (UF 20892), a cervical vertebra 
(UF 209OA), and a supraocci p i tal crest (UF 20905) represent 12 to 20 
individuals ranging from 65 mm to 195 mm carapace length (CL) ; many 
additional elements from this deposit, particularly plastral and 
peripheral bones which lack diagnostic features, probably represent 
D. reticutaria as well; fossils from the site are indistinguishable 
from modern D. retioularia, their shallow rugosity probably reflect- 
ing their relatively small size; STI 0.95 to I.05. 

Warm Mineral Springs: To date, 35 elements - one nuchal, seven neural, 
one suprapygal, six pleural, 13 peripheral, and five plastral bones, 
plus a scapula and broken femur - all assigned field number WMS 19352 
and representing five to ten individuals of CL 1 38 to 18^4, have been 
removed from this site. The bones are similar to those from Nichoi's 
Hammock and have an average ST! of 1.15- 

Vero: A large number of plastral and carapacial elements, including 
at least two nuchal, two neural and two pleural bones (all recently 



26 



acquired by the Florida State Museum as part of the former Florida 
Geological Survey collection and as yet uncatalogued) are virtually 
indistinguishable from modern D. reticularia; ST! O.85-O.95. 

Waccasassa River I: Two second neural bones (TRO 101, 102) and a third 
neural bone (TRO 103), representing three individuals of I30 to 210 CL 
(Fig. 9); STI 1.1 to 1.3- 

Waccasassa River V: A lightly-sculptured nuchal bone, UF I627I (Fig. 
9): greatest length 30.5, greatest width 35-5, estimated CL 135; 
proximal end of a pleural bone, UF 16275; STI 1.1. 

Waccasassa River VI: A distinctly grooved nuchal bone, UF 2I906: 
greatest length 39-8, greatest width 42.3, estimated CL 170; STI 1.1. 

Reddick IIC: A first neural bone (UF 21955) from an adult turtle 
(estimated CL I80) and the proximal end of a fourth pleural bone from 
a juveni le; STI 1.1. 

Coleman IMC: Four elements (UF I5I86E) representing at least three 
individuals: a longitudinally rugose, relatively deeply notched pygal 
bone (length 21.5); a left epiplastron ( interep i plast ral suture length 
13.6); a characteristically rugose left xi ph i pi ast ron missing its distal 
portion (hypo-xiphiplastral suture length ^40. 7); and a distinctly 
sculptured right hypoplastron ( interhypoplast ral suture length 58.4); 
STI 1.3. 

St. Petersburg, Catalina Gardens: Lower two thirds of a right fifth 
pleural bone (UF 19248): greatest width 30.0, estimated length 60, 
estimated CL 220; STI 1 .3. 



27 



Seminole Field: A deeply sculptured fragment of a right second pleural 
bone (UF 9927) with rib attachment - width at rib level 28.0, thickness 
at rib level 5-9, estimated CL 210; fragment of a left hypoplastron 
(UF 9927) with deep longitudinal grooves on ventral surface, estimated 
CL 210; STl 1 A. 

Bradenton 5!st Street: A characteristically sculptured fragment of a 
nuchal bone (UF 2482): estimated CL 210, STl 1.25. 

Kendrick lA: A sixth neural bone (UF 19250) with a pronounced, scale- 
like sculpturing and a low, rounded keel - greatest length 19.3, greatest 
width 33.2, greatest thickness 6.0, estimated CL 250, ST! 1.3; a deeply 
grooved partial nuchal bone (UF 9292) possibly from the same individual 
and described previously by Jackson (1964): estimated CL 250, STl l.I 
to 1.6; (Fig. 10). 

Wall Company Pit: Proximal halves of two broken pleural bones (UF 5026): 
a second left (estimated CL 175, STl 1.6) and a deeply rugose fourth 
right (estimated CL 220, STl I.5) with rib distance: pleural width 
ratios of 0.84 and O.8O, respectively. 

Haiie XVI: 38 elements representing at least 15 individuals of CL 116 
to 240: a nuchal bone (UF 20896) , length 40.0, estimated CL 182; a 
second neural bone contiguous with the second and third right pleural 
bones (UF 20888; Fig. 11), and the first left and second right periph- 
eral bones (UF 20889) almost certainly from the same individual, 
estimated CL 230; fifteen fragmentary pleural bones (UF 20895; UF 20898) 
and seven peripheral bones (UF 21970); a hypoplastron (UF 2I969) and 



28 



partial hypoplastron (UF 21968); contiguous second, third and fourth 
neural bones (UF 20893) from a turtle of 227 CL; and the third 
(UF 20897), fourth (UF 21971), two fifth (UF 20894 and UF 20898) , and 
sixth (UF 20898) neural bones from five turtles with CL of 240, ]kO, 
225, 160 and 220, respectively. Neural length: width and rib distance: 
pleural width ratios are included in Figs. 7 and 8; STI of neural 
bones 2.0 to 2.2. 

Haile XV: A fifth neural bone (UF 192^9), the dorsal surface of which 
is extremely flat but moderately sculptured: greatest length 20.2, 
greatest width 32.0, estimated CL 210; an anterior fragment of a nuchal 
bone (UF 19168), estimated CL 230; STI 1.5; (Fig. 12). 

Discussion of fossil material . All of the Ranchol abrean and 
Subrecent material is clearly referable to D. retiaularia. With the 
exception of shell thickness, relative dimensions of individual fossils 
show no significant differences from corresponding measurements of extant 
turtles. The Blancan and Irvingtonian material, as well as the Kendrick 
nuchal, indicate that this species reached a slightly larger maximum 
size during the Late Pliocene and Pleistocene than at present. The 
blunt median keel on the Kendrick neural, although not typical of most 
extant D. reticulavia, occurs posteriorly in a few individuals. Though 
tending to be more pronounced in the Pleistocene, shell rugosity 
patterns are within the range of variation of modern B. veticularia. 

The single consistent difference between Pleistocene and Recent 
D. veticularia is that of shell thickness. The STI of Pleistocene D. 
veticularia is I.l to 2.2 times that of Recent turtles. The trend 



29 



towards shell thickness reduction appears roughly chronoclinal since 
at least the Irvingtonian (Table 2), though the absence of material 
from some glacial and interglacial periods may conceal unseen fluctua- 
tions. Similar trends in post-Pliocene shell thickness reduction have 
been suggested, though less well supported by a t ime-t ransgress i ve 
series of fossils, for Chvysemys (Preston, I966, 1971), Emydoidea 
(Taylor, 19^3), Graptemys (Chapter l)^ Kinosteimon (Fichter, 1 969), 
Trionyx (Wood and Patterson, 1973) and Geochelone (Auffenberg, 1963b). 
Although shell thickness alone is inadequate as a basis for taxonomic 
separation, it is not a simple function of turtle size as Auffenberg 
(1958) states for Terrapene. From Middle Pleistocene to the present, 
the shell of D. reticularia has become progressively thinner. Gaps 
in the STI-time curve may reflect our incomplete sampling of the 
fossil record. Nevertheless, the possibility of sexual and ontogenetic 
polymorphism in this character, as well as hidden fluctuations in the 
curve, could complicate the matter. Unfortunately, sample sizes from 
most fossil sites are inadequate for a thorough treatment of the data. 

The Irvingtonian (Haile XVI) fossils differ from younger material 
in two additional ways: a higher length: width ratio for the second and 
third neural bones (Fig. 7) and a slightly more proximal site of rib 
juncture with the second and third pleural bones (Fig. 8). These 
characters do not exhibit allometry in Recent adult turtles, and there 
is thus no reason to suspect it in Pleistocene populations. In these 
respects Irvingtonian Deivochelys are morphologically intermediate 
between the later Pleistocene and the Middle Pliocene turtles discussed 
below. The near identity of the Blancan neural (UF 192^9) with that of 



30 



Table 2. Shell thickness index (STI) of fossil Deirochelys 
from 16 Florida sites listed chronologically by 
faunal periods. 



Age and Site 


STI 


Age and Site 


STI 


Ari kareean 




Rancholabrean, cont. 




Thomas Farm 


1.9 


Bradenton 


1.25 


Hemp hi 1 1 ian 




Kendrick 


1.3 


Love 


1.6-2.1 


Semi nol e Field 


\.h 


Mixson 


1.8 


Catal ina Gardens 


1.3 


Haile VI 


1.9 


Coleman IMC 


1.3 


Blancan 




Waccasassa 1 


1.1-1.3 


Haile XV 


1.5 


Waccasassa V 


1.1 


1 rvington ian 




Subrecent 




Haile XVI 


2.0-2.2 


Warm Mineral Spring 


1.15 


Rancholabrean 




N ichol ' s Hammock 


0.95-1.05 


Wal 1 Co. Pit 


1.5-1.6 







31 



an Irvingtonian one (UF lOSSk) suggests that little shell evolution was 
experienced between these periods. 

The differences between modern and Upper Pliocene to Middle Pleisto- 
cene Deiroohelys are real and might justify taxonomic distinction were 
it not for the intermediate Ranchol abrean material. Such time-related 
changes are, however, to be expected within a chronocllnal lineage, 
as shown previously by Mil stead (I967) with Tevrapene. The modern 
subspecies of D. veticulavia are distinguished by coloration and shell 
shape (Schwartz, 1956) and consequently can not be compared to these 
fossils. Furthermore, D. retioularia likely varied geographically 
during the Pleistocene as it does now. For these reasons I refrain from 
erecting subspecific epithets for any of the Pleistocene or Upper Plio- 
cene fossils and simply refer them all to the species Deirochelys 
retiaularia. 

The existence of certain morphological differences, reflected in 
carapacial osteology, of specimens referred above to D. veticulavia and 
all earlier representatives of the genus (Figs. 7 and 8) is accentuated 
by the absence of Late Hemphill ian fossils. This gap in a gradually 
evolving lineage creates a convenient (though admittedly artificial) 
point of division between morphologically distinct forms. I therefore 
designate the turtle represented by the Middle Pliocene fossils as 

Deivochelys cavvi new species 

Etymology . Named in honor of Archie F. Carr for his extensive 
contributions to our knowledge of Recent turtles and to herpetology in 
general . 



32 



Holotype . UF 20908, a fragmented but nearly complete carapace 
lacking only the nuchal bone, first neural bone, and anterior peripheral 
bones (Fig. 13A); a partial plastron consisting of the left hyoplastron, 
hypoplastron, and x iphi pi ast ron apparently represents the same individua: 
(Fig. 13B). 

Type locality and horizon . Alachua Clay, Love Bone Bed, near 
Archer, Alachua County, Florida, Early Hemphill ian. Middle Pliocene. 

Referred material . All from four Florida sites producing 
Hemph i 1 1 ian faunas : 

Mixson's Bone Bed: a fourth neural bone, UF 20890 (formerly Florida 
Geological Survey V-2599) , assigned previously by 0. P. Hay (1916) to 
Chvysemys aaelata: estimated CL 290, STI 1.8. 

McGehee Farm: a complete (UF 1920^) and two partial (UF 2089I and UF 
20903) nuchal bones - measurements of UF 19204: length 57.1, width 
60.8, corresponding to a CL of approximately 263; right hypoplastron, 
ninth right peripheral bone, and left and right x i ph i pi astral fragments 
(UF 20899). 

Haile VI: contiguous second neural bone fragment and proximal portion 
of left second pleural bone, (UF 20887); estimated CL 253, STI 1.9; 
contiguous pygal bone and eleventh left peripheral bone, (UF 6485a); 
anterior end of third cervical vertebra (lacking zygapophyses) , UF 
6485b (Fig. 14); five peripheral bones, UF 6485c; first neural bone, 
(UF 6485d); seventeen pleural bone fragments, (UF 6485e) ; many other 



33 



elements and fragments from this site may represent either D. carri or 
Chrysemys caelata (Chapter ill). 

Love Bone Bed: Although excavation is incomplete, this deposit is already 
the richest source of fossil deimohelijs known. At the time of this 
writing over ^400 carapacial elements and half as many plastral elements 
of Beivochelys have been removed. Other than the holotype, only two 
sets of associated carapacial bones have been found (UF 2^100 and UF 
20900, Fig. 15). The less water-worn carapacial elements display the 
distinct scale-like sculpturing characteristic of the genus (Fig. 16). 
Many elements represent turtles of exceptionally large size for Beivo- 
chelys: the largest nuchal bone (UF 20906) measures 59.8 (length) x 
61.8 (width). The average STI range is 1.6 to 2.1. 

Diagnosis . Deivochelys aavri differs from D. veticularia in having 
relatively narrower neural bones (mean length: width ratio of third 
through fifth neural bones 0.8 to 0.9; Figs. 7, 13A) and a more proximal 
site of emergence of the ribs from the pleural bones (Figs. 8, 15). 
Elongation of cervical vertebrae and patterns of shell rugosity are 
similar in these species, but the carapace of D. carri appears to be 
relatively broader. 

Shell rugosity and width of first vertebral scute of D. carri 
are like those of Chrysemys caelata and C. williamsi , respectively, also 
from the Florida Pliocene (Chapter III); nevertheless, other generic 
characters distinguish these species from Z). carri. Neural bones of 
B. carri are similar in shape to those of the Florida Pliocene Chrysemys 
inflata (Weaver and Robertson, 1967), yet distinguished from them by 



31^ 



absence of the pronounced keel and deeply excavated surface of the 
latter. 

Description . With the exception of the less developed character 
suite previously alluded to, D. cavvi is, in most respects, similar to 
B. veticulccpia. Nevertheless, many of the fossils indicate that the 
former reached a greater size than B. reticulavia, perhaps as large as 
320 mm CL, compared to approximately 250 mm CL today (Carr, 1952). The 
shell of D. carri is about twice as thick as that of extant D. retioularia 
but not unlike that of Blancan and Irvingtonian representatives of the 
modern species (Table 2). Additionally, the reconstructed holotype 
shell is relatively broad and flat compared to Recent chicken turtles. 
In this respect, as well as in the flaring of the posterior peripheral 
bones, D. carri is reminiscent of some members of the genus Chrysemys 
and appears to have been more streamlined than D. reticularia. One 
fairly constant difference between D. reticularia and D. carri is that 
the anterior edge of the fourth vertebral scute (incised at the fifth 
neural bone) of D. reticularia projects forward to form a sharp 
anteriorly-directed V, whereas that of D. carri projects forward only 
slightly (and more bluntly) or not at all (Fig. 1 3A) . The plastron of 
V. carri, like that of D. reticularia, is narrow. The anal notch in the 
plastron associated with the holotype of D. carri is twice as deep as 
that of D. reticularia. There is no significant morphological variation 
among D. carri from the four sites. Measurements and qualitative obser- 
vations of all material from Halle VI, McGehee Farm, and Mixson's Bone 
Bed fall within the range of variation of elements from the Love Bone 
Bed. 



35 



Discussion. Veirochelys oarri is similar in most respects to its 
presumed descendant D. veticulavia . The major differences are modifi- 
cations associated with the further development of the specialized 
elongate neck and head in D. reticularia. In this respect both D. 
cavvi and B. veticulavia surely represent segments of a single chrono- 
clinal lineage. The neural spines and dorsal rib heads of B. cavvi 
are typical of the genus and only slightly more robust than those of 
D. veticulavia. The single cervical vertebra (UF 6^85b) referable to 
D. cavvi (Fig. U) is likewise slightly more robust than the correspond- 
ing vertebra of D. veticulavia; additionally the posterolateral flanges 
on the modern centrum, which must serve as muscle attachment surfaces, 
are lacking in the fossil (the possibility of wear is unlikely). 
Although it is impossible to determine accurately the length of the 
Pliocene vertebra from the Haile VI fragment, it appears that the 
characteristic cervical elongation and development of associated 
modifications in Deivochelys had already approached present levels by 
Middle Pliocene. Nevertheless, D. cavvV s narrower neural bones and 
more proximal rib union with the pleural bones relative to D. veticulavia 
imply a shorter free rib between the pleurals and vertebral column and 
a correspondingly less developed set of cervical extensor muscles in 
the former. A slightly shorter or less powerful neck in the Pliocene 
species therefore seems likely. Certainly any future finds of Deivo- 
chelys skull and cervical material in the Love Bone Bed would be 
particularly valuable. 

Although the Love Bone Bed provides us with an exceptionally 
fine series of Deivochelys fossils, far older than any previously known 



36 



for the genus, we can trace the evolutionary record of this turtle back 
still one step further - to the Miocene. 

The Thomas Farm Deivochelys 

The only emydine turtle previously recognized from the Florida 
Miocene (Thomas Farm) is a species Chrysemys of uncertain status (Williams, 
1953; Rose and Weaver, I966). In an effort to determine the relation- 
ships of this turtle, I examined the holdings of the Flroida State 
Museum for additional material. Among the elements retrieved were a 
faintly sculptured neural bone (UF 219^+9) only slightly narrower than 
those of D. carri , and the proximal fragment of a pleural bone (UF 
21950) with a rib juncture scar too low for that of Chrysemys (Fig. 
17)- Comparisons with Recent and fossil Deivochelys and Chrysemys , 
including "typical" Chrysemys elements from Thomas Farm, leave no doubt 
that the two fossils represent Deiroohelys. Curvature of the scute 
sulcus, relative length of the anterolateral borders, and extreme 
lowness of the neural spine all indicate that the neural bone is a 
fifth, while the relative proportions of the medial borders of the 
pleural bone in addition to the position of the sulcus indicate that 
it is probably the second pleural bone from the left side. As with 
the Pliocene Deiroohelys , the shell is relatively thick (STI, 1.9). 

In addition to the two fossils described above, I tentatively 
refer to Deiroohelys the following elements from Thomas Farm: one 
complete epiplastron (UF 21932) and the medial half of another (UF 
21939), the posterior part of a right x i phi pi astron (UF 219^6), the 
major part of an entoplastron (UF 219^2), and the proximal end of a 



37 



pleural bone (UF 21951). Additionally, one complete and two fragmentary 
nuchal bones (MCZ 3^32; see Fig. 4 in Williams, 1953, and Fig. 2B in 
Rose and Weaver, 1966), although probably representing Chrysemijs, may be 
Deiroohelys. The width of the first vertebral scute and shape of the 
nuchal scute are like those of both Deiroahelys and CJirysemys ovnata. 

Both the shape of the neural bone (length: width ratio, 0.94, 
Fig. 7) and the point of juncture of the rib with the pleural bone 
(rib distance: pleural bone width ratio, Q.k], Fig. 8) indicate that, in 
terms of cervical hypertrophy, the Thomas Farm Deiroahelys was even more 
primitive (less specialized) than D. cavvi. Remains of the very low 
neural spine fused to the neural bone confirm this. Hence, I believe 
that the limited Thomas Farm material represents a turtle distinct from 
B. cavvi. However, any taxonomic assignment of the Thomas Farm fossils 
other than to genus must await additional and preferably associated 
material. More important at present is that in the Thomas Farm Miocene 
we find an important link in the gradual evolutionary sequence from a 
generalized emydine ancestor (cf. Chvysemys) into the more specialized 
d. cavvi and its highly specialized descendant, d. vetioulavia. 

D i scuss ion 

The material now available shows that the genus Deivochelys, 
instead of being an evolutionary enigma, possesses possibly the most 
complete evolutionary record of any modern turtle genus. Evolution of 
Deivochelys has been by specialization of a generalized emydine stock 
(presumably Chvysemys). The earliest fossils are, in fact, difficult 
to distinguish from Chvysemys. We may estimate by extrapolation at 



what point the two genera would be no longer d i st inct-- i . e. , the time 
at which a generalized turtle began its Initial shift to a new adaptive 
zone in response to selective pressure. The elongated neck (and pre- 
sumably skull) as well as associated muscular (Shah, I963) and osteo- 
logical modifications of Deivoohelys had already developed by Middle 
Pliocene. This character suite is already conspicuous In hatchl ing 
D. veticularia, so that phylogenetic recapitulation must occur very 
early during ontogenetic development if It occurs at all. The divergence 
from a more generalized aquatic emydine stock (moderately short neck, 
long neural bones, weak hyoid apparatus, robust ribs emerging from very 
near the proximal ends of pleural bones, limited trunk vertebral 
musculature, and a relatively broad shell, as In the genus Chvysemys Br^A 
some members of the Eocene genus Echmatemys) had certainly begun by the 
Miocene. Extrapolations based on an average rate of evolution from such 
a generalized ancestor suggest an 01 igocene origin of the genus (Fig. 
18). This character suite almost certainly evolved as a peculiar 
trophic structure; Deivoohelys utilizes a "pharyngeal" method of feed- 
ing (Bramble, 1973) for capturing prey capable of quick movements (pri- 
marily aquatic arthropods). Arguments such as those of Webb and 
Johnson (1972), In which cervical elongation Is held to represent a 
thermoregulatory device, seem at most of secondary significance in 
this case, particularly in light of the hypoertrophi ed hyold skeleton. 

The thick shell of D. cavvi and the Thomas Farm deivoohelys, as 
well as of Blancan and Irvingtonian B. vetioulavia (Table 2), suggests 
that until Late ?\&\stoc&n&, Deivoohelys was a moderately thick-shelled 
turtle. Pleistocene reduction in weight and volume of the shell may 



39 



have allowed faster pursuit and increased maneuverability necessary for 
capturing fast-moving prey (author's unpublished data) on which Deiro- 
ohelys had come to specialize. Loss of armor (if the thick shell served 
this purpose) may have been offset by crypsis and behavioral immobility 
(unpublished observations). In addition to changes in shell thickness, 
general reduction in body size, accompanied by relative elongation and 
heightening of the shell, seems to have occurred from at least Hemphillian 
to Rancholabrean times. 

Rel at ionshi ps 

Baur (1889) was the first to hypothesize a close relationship be- 
tween Emydoidea and Deirochelys on the basis of similar skull and rib 
specialization. Although Carr (1952) believed the similarity between 
Emys (= Emydoidea) hlandingii and D. veticularia to be "purely 
fortuitous," most subsequent workers supported Baur's idea. Bramble 
(197^) summarizes the situation: 

Williams (in Loveridge and Williams, 1957) presented a 
forceful case for a relationship between Emydoidea and 
Deirochelys. Although DeivooheZys possesses no plastral 
hinge and on many points of shell morphology closely 
approaches certain members of the genus Chrysemys 
(McDowell, 196^), it does, as Williams noted, share with 
Emydoidea a number of specializations of the skull, 
cervical vertebrae and neck musculature. On these 
grounds Williams suggested that Emydoidea was a derivative 
of Deirochelys and only convergent with Emys. This view 
has been widely adopted by later workers (Tinkle, 1962; 
McDowell, 1964; Zug, I966; Pritchard, 1967; Milstead, 
1969; Ernst and Barbour, 1972), some of whom (Tinkle, 
1962; Zug, 1966) have presented additional evidence In 
support of it. McDowell (1964: 275) found no 'signifi- 
cant cranial differences between Deirochelys and Emydoidea^ 
and accordingly placed both genera in a Deirochelys Complex 
within the Emydinae. (p. 12h) 



ko 



However, Bramble's (197^) study of shell kinesis and other osteological 
and myological characters indicates instead that Emydoidea is a "close 
phyletic associate of 5'mys and Tepi^opene" as well as of Clenmys (the 
four genera comprising the Ctemmys Complex), and that these genera may 
be distinguished as a group from Deivochelys and McDowell's (1964) 
Chrysemys Complex. Waagen (1972) formed an identical opinion from 
his analysis of musk glands in Recent turtles. On the basis of fossils 
discussed in this paper I agree with the conclusions of Waagen (1972) 
and Bramble (197^) that Deiroahelys shares a close relationship with 
the genus Chrysemys, and that similarities hetvieen Emydoidea and Deiro- 
ahelys are "undoubtedly the result of convergent feeding system" 
(Bramble, 197't). In fact, most of the modifications used to 
substantiate a close relationship between Deivochelys and Emydoidea 
(elongated ventre 1 1 y-bowed free ribs, widened neural bones, elongated 
cervical vertebrae, and a greatly hypertrophied cervical musculature) 
are also present in the totally unrelated (at least at the familial 
level) cryptodire genus Chelydra as well as the pleurodire genus 
Chelys. They are, moreover, all modifications associated with the 
pharyngeal method of feeding (Bramble, 1973) employed by these turtles. 
Hence, the taxonomic use of this particular character suite, so clearly 
convergent among members of three distinct families, should be treated 
cautiously in attempts to determine i ntrafami 1 ia 1 relationships. This 
paper has presented evidence of the gradual development of these 
adaptations as a unit of functional morphology (Wilson, 1975) within 
one of these phyletic lines. 

Pleistocene and Late Pliocene fossils of Emydoidea, which are 
clearly referable to the modern species E. blandingii (Taylor, 19^*3; 



^1 



Preston and McCoy, 1970, show no special resemblances to Late Tertiary 
Deirochelys , other than the convergent character set already discussed, 
and hence do not support a theory of their divergent evolution. Neither 
the fossil records nor the present distributions (Carr, 1952; Preston 
and McCoy, 1971; Zug and Schwartz, 1971; McCoy, 1973; Jackson and Kaye, 
197^) give any indication that the two genera were ever sympatric, 
although the southern extension of the range of Emydoidea in the Late 
Pleistocene (Jackson and Kaye, 197^*) closely approaches the present 
northern limit of Deirochelys in Mississippi. Further ecological 
studies might help to determine if this allopatric relationship reflects 
a Gause-type competitive relationship or a difference in thermal re- 
qu i rements . 

Distribution and Paleoecology 

The genus Deirochelys is endemic to the southeastern United 
States, and it is therefore not surprising that the first extensive 
evidence of its fossil history should be from Florida. All vertebrate 
fossil sites known to contain Deirochelys (Fig. 3) occur within the 
range of the modern subspecies D. reticularia chrysea or its zone of 
intergradat ion with D. r. reticularia (Schwartz, 1956; Zug and 
Schwartz, 1971). 

Deirochelys reticularia usually inhabits quiet, shallow bodies 
of freshwater throughout its range although it occasionally enters the 
quieter portions of streams (Pope^ 1939; R- Webb, 1950; Carr, 1952; 
Schwartz, 1956; Campbell, I969) and perhaps rarely saltwater (Neill, 
19^*8; Martof, I963). Personal observations in north-central Florida 



k2 



indicate that the densest populations of Deivochelys occur in shallow 
(less than one meter) ponds with abundant basking logs, emergent bushes 
(e.g., Cephalanthus), and an extensive Lenma-Wolffiella surface mat. 
From a structural standpoint, the relatively short limbs, long nuchal 
scute underlap, and absence of streamlining (as compared to a lotic 
form such as Chrysemys conoinna) reflect its evolution as a quiet- 
water form. The turtle also shows a proclivity for overland wanderinc 
(Neill, 19^*8; Carr, 1952; Gibbons, I969, 1970). Its typical association 
with the Southeastern Coastal Plain (Mount, 1972; Mount and Folkerts, 
1968) implies adaptation to a warm temperate climate. The presence of 
Deivochelys and associated fauna {he-pisosteus, Amia, Alligator, Chry- 
semys aaelata {Chapter III), Trionyx cf. T. ferox] in Hemphillian sites 
thus indicates the existence of quiet freshwater (e.g., sinkhole 
ponds or sluggish streams) and a warm, equable climate in the Florida 
Midd 1 e PI iocene. 

Even In the most favorable habitats Deivochelys today rarely 
reaches densities comparable to those of sympatric emydine turtles (e.g., 
Chrysemys nelsoni, C. flovidana, C. scvipta) . This relationship appears 
to hold also in the Pliocene; in the only Pliocene deposit containing 
large numbers of Deivochelys (Love Bone Bed), Chrysemys caelata elements 
outnumber those of D. carri approximately four to one. In what presumably 
was a suboptimal habitat for Deirochelys at McGehee Farm the ratio is 
even more disparate. This indicates that populations of Deirochelys 
may be more restricted by limiting factors than are other emydines. 

All fossil records for Deirochelys from sites near the present 
coastline of Florida (Fig. 3) are either Subrecent or Late Ranchol abrean. 



^3 



All other sites except those in the Waccassassa River are in presently 
wel 1 -drained localities 21 to 37 m above present sea level; these in- 
clude all sites assigned to the Hemphillian, I rv ington ian , Blancan, 
and early Rancholabrean periods. 

Webb and Tessman (I968) have presented vertebrate faunal evidence 
supporting conclusions based on physiographic evidence (Alt and Brooks, 
196^; Alt, 1967) that sea level dropped and rose again as much as 30 
meters during Hemphillian (middle Pliocene) time. McGehee Farm (early 
Hemphillian) was thus very near the Pliocene coastal shoreline during 
its time of deposition and its fauna clearly reflects an estuarine 
influence, although nearby Mixson's Bone Bed, which occurs at the same 
elevation, does not (Webb, 1964; Webb and Tessman, I968). Additionally, 
the Late Pliocene and Middle Pleistocene interglacial deposits contain- 
ing Deivochelys were much nearer to coastal shorelines during deposition 
than they are today. It seems probable that since at least the Plio- 
cene Deivochelys has been associated primarily with lowland habitat, 
as was the Pleistocene box turtle subspecies Tevrapene cavolitia putnami 
in Florida (Auffenberg, 1958). Distribution of these turtles in the 
Florida peninsula and along the Gulf coast must have fluctuated with 
the advance and retreat of the Pleistocene sea. The proclivity of the 
genus for overland wandering has probably been instrumental in maintain- 
ing or reestablishing inland populations at higher elevations in the 
abundant "perched" lakes (bodies of water which are completely above 
the piezometric surface and sometimes subject to spontaneous drainage) 
common throughout much of the peninsula today. The present inland 
populations may be relicts of higher sea levels or terrestrially- 
reestablished populations. 



Figure 3- Fossil sites in peninsular Florida containing Deiroohelys. 
Site ages are given in Appendix. 

9. Semi nole Field 

1 0. Catal i na Gardens 

1 1 . Bradenton 

12. Warm Mineral Springs 

13- NIchol's Hammock 

]k. Reddick I IC 

15. Thomas Farm 

16. Vero 



1 


McGehee Farm 


2 


Ha i le si tes 


3 


Love Bone Bed 


h 


Wal 1 Company P i t 


5 


Mixson's Bone Bed 


6 


Kendrick lA 


7 


Waccasassa River sites 


8 


Col eman IMC 



'f5 







Figure k. Third neural bones of Chrysemys floridana (A) and 

Deivoahelys vetiaularia (B) ; note greater width and 
characteristic sculpturing of latter. 



Figure 5- Frontal aspects of third pleural bones of Chvysemys 

aoncinna (A) and Deivoahelys vetiaularia (B) , showing 
dorsal ribs. 



if? 



/ 






o 


> 


:,■■ 


1- 






',- 


"O 


» 




Q) 


r^'^ 


-o 


4-1 


I 


rz 


m 


') 


11' 


— 


w 



^9 





Q 





Figure 7. Length: width ratios of second through sixth neural 

bones of Recent (closed circles), Irvingtonian (stars), 
Hemphillian (triangles - Love; square - Haile VI; 
asterisk - Mixson's), and Arikareean (open circle) 
Deiroohelys. Dice - Leraas diagrams depict mean, 
range, and two standard errors; numbers above and below 
bars represent sample sizes. 



51 



1.0-1 



10 



10 



1 
O 



10 



X 

9 

I— 
o 
z 

LU 



< 

OH 

LU 

Z 



f 



10 







1 


i 


L 





1 

• 



.6- 



* 



10 



10 



10 



—r 
4 



-I- 
6 



NEURAL NUMBER 



Figure 8. Rib distance: pleural bone width ratios for Recent, 
Pleistocene, and Pliocene Deiroahelys; all symbols 
as in Fig. 7- 



53 



l.Oi 



X 

a .8 



< 



'^ .6 

LU 
U 

z 

to 
Q .4 

CO 



.2' 






4 f 




tti tl 



f 



3 4 5 

PLEURAL NUMBER 



6 



Figure 9. Nuchal (UF 16271) and three neural bones (TRO 101-103) 

of Rancholabrean Deivoohelys reticularia from Waccasassa 
River V and I, respectively. 



Figure 10. Distinctly sculptured nuchal bone (UF 9292) and sixth 
neural bone (UF 19250) of Deirochelys reticularia 
from Kendrick lA. 



55 



^^^^R^, «Pp 



w 





Figure 11. Dorsal (A) and ventral (B) surfaces of carapace fragment 
(UF 20888) of Irvingtonian Deirochelys retiaularia 
(Haile XVl). Note sculpturing, neural bone width, 
and rib junctures. 



Figure 12. Fifth neural bone (A, UF 192^9) and nuchal bone fragment 
(B, UF 19168), of Blancan Deirochelys retiaularia (Haile 
XV), showing broad nuchal scute underlap. 



57 





• s. 



%■ 



V^' 



•f\ 




Figure 13A. Beirochelys carri holotype, UF 20908, dorsal aspect 
of carapace. Hatched areas missing from fossil. 



59 




Figure 13B. Deivochelys oarvi holotype, UF 20908, ventral aspect of 
plastron. Hatched areas missing from fossil. 



61 




!^ 


, — 


c 


C^ 


(D 


>— 


Vl 


I- 






D 


1- 


03 


OJ 


3 


;^r1 


1 — 


o 


r-i 


o 


o 


W 




o 


r^' 


JZ 




n 


4J 


■o 


(1 


•— 


r— 


Is 


i 


3 


•vJ 




f) 


W 


lA 


? 


t^ 


<u 





— c x: 0) 



— « 

— « 
> t~^ 

0) o 
O M 



OJ 


"S 




03 


•t-l 


o 


(U 


> 


if) 


x: 


•M 




o 


tf) 


(D 


3 


o. 




U 


tr 




-. 


• — 


0) 


M- 


o 


■a 




o 


o 


r 


4- 




CTl 


>— 


o 


(U 


o 






o 


CM 


in 


fi 


(D 




X 


•t^ 


M- 




o 


?., 


1- 


-3 


1- 


(-i 


3 




1- 


r-i 


in 


•^ 


< 


;^; 




OJ 




i:^ 


f— 


u 




•vJ 


m 


(0 


- 


-(^ 


i_ 


Q. 


</l 


w 


4-> 


TO 


<u 


Sh 


c 


u 


r 




<u 


01 


o 




> 


o 


-D 


t-:i 



63 





Figure 16. A distinctly sculptured nuchal bone (A) and posterior 
peripheral bone (B) of Deiroohelys oarri from the 
Love Bone Bed (x 1.15). 



Figure 17. Beirochelys fossils from the Thomas Farm Miocene 
(X 2.25). (A) Neural bone, (B) visceral surface 
of pleural bone fragment showing rib juncture scar. 



65 





B 





B 





o 


— 


<u 


• — 


T> 


c 


■M 


>• 


o 


<u 


h 


X! 


-C 


<u 




■l-J 




-— 


() 


TD 


01 


ex 


(U 


!_ 


>- 


N 


13 


J= 


— 



o 


0] 


0) 


o 





m 


<u 


• — 




in 


TJ 


<u 




C 


i- 


14- 


. — 





U 


„ 


b 


0) 


CQ 


ro 


f- 


;s) 




— 


!-^ 


£ 


■I-I 


w 







rS- 


i_ 


JJ 


V^ 


M- 


Dl 


n 




3 


?H 


<U 



O I- o 

CL 

(D -C -I-I 

1- 4-> C 0) 

4-1 -O .- C 

X — O — 

Lu 3 a. .— 



67 





(/> 


CO 


>: 


>- 


LU 


^ 


X 


UJ 


u 


to 

>- 


o 


o^ 


Q^ 



iN 



U 



I / 
I / 
I / 
1/ 



c 

u 
O 



C 
Q) 
U 
O 





C 

u 
O 
O) 



o 



m 



o 



"T" 



q: 

00 00 

to 

< 



-O 



to 

z 

O 



CN 



M : 1 ivyn3N aN3 



CHAPTER I I I 
THE STATUS OF THE PLIOCENE TURTLES 

CHRYSEMYS CAELATA (HAY) AND CHRYSEMYS CARRI ROSE AND WEAVER 

Three non- Trachemys species of Chrysemys (sensu McDowell, 196^) 
have been described from the Pliocene ofRorida. Hay (I908) described 
C. caelata, largely on the basis of shell sculpturing, from MLxson's 
Bone Bed," Levy County. Hay considered this site Pleistocene in age 
although it had previously been recognized as Pliocene (Dall and Harris, 
1892; Leidy and Lucas, I896) and was so reassigned by Simpson in I929. 
Rose and Weaver (I966) examined shells of Chrysemys from McGehee Farm 
in adjacent Alachua County and described both a smooth-shelled species 
{C. williamsi) and a rugose species {C. carri) yet made no reference to 
C. caelata. Both sites are Pliocene deposits within the Alachua 
Formation (Simpson, 1929a; Rose and Weaver, I966; Hirschfeld and Webb, 
1968). The occurrence of two rough-shelled species in approximately 
equivalent strata only 3k km apart prompted me to investigate their 
taxonomic status. All specimens are in the collections of the U.S. 
National Museum (USNM) , Florida State Museum (UF), and Florida 
Geological Survey (FGS). 



"Uncertainty surrounds Hay's designation of the type locality of C. 
caelata, as it does much of the early Mixson material (Simpson, 1929a). 
In conjunction with the original description. Hay (I908) gave the 
locality as '"Mason's bone bed,' somewhere in Levy County, Florida," 
but later (I9I6) referred the site only to "somewhere in Levy County 
Florida," The data with the USNM specimens read "Levy County, Florida 
1885, L. C. Johnson." As this is the period when Johnson and other 
representatives of the U.S. Geological Survey were originally 
investigating Mixson's Bone Bed (Simpson, 1929a), and as the material 
exhibits the same condition of preservation as known Mixson material 
I suggest that "Mason's" is a liberal orthographic interpretation of 
Mixson' s." 



68 



69 



Taxonomic Considerations 

Chrysemys caelata was described by Hay (1908) on the basis of 11 
unassociated elements; of these he considered 10 (USNM 2508) as probably 
representing one large individual and a single pleural (USNM 6064) a 
smaller specimen. The nuchal bone was designated as the type. My 
examination of this material reveals that probably no more than two bones 
represent any single individual. However, the entire series appears to 
represent one species. The two posterior peripherals to which Hay 
(1908) referred are a tenth and eleventh from the right side. In his 
original description (I908) he wrongly referred to the seventh right 
peripheral as a third left, but corrected this error in 1916. As Hay 
gives adequate measurements and detailed accounts of the sculpturing on 
most of the bones, a redescript ion of the material is unnecessary. 

Hay (1908) placed great emphasis on the characteristic sculpturing 
of the shell of C. caelata, which he said "resembles that of Traahemys 
saripta" (= Chrysemys) . The sculpturing bears an even stronger 
resemblance to that of C. nelsoni, which, 1 i l<e saripta, occurs in 
Florida today. Having only limited material from which to describe 
the sculpturing. Hay did not appreciate the range of i ntraspeci f ic 
variation to which this character is subject. Other specified diagnostic 
characters include a wel 1 -developed epiplastral lip and absence of 
nothces along the posterior carapace edge. 

Rose and Weaver (1966) described Chrysemys oarri from a series of 
incomplete shells (holotype UF 9^27 and nine paratypes). They diagnosed 
it as "a species of Chrysemys with a rugose carapace and plastron, a 
slight median posterior keel, long nuchal scute underlap..., moderate 



70 



gular scute overlap..., un-notched peripherals," large epiplastral lip, 
"and a clearly notched and rectangular pygal bone." 

I have examined and compared the type series of both C. oarri 
(except UF 1 1 083 which could not be located) and C. caelata. Patterns of 
surface rugosity and scutellation of the individual elements of C. caelata 
are well within the range of and often virtually identical to sculpturing 
and scutellation patterns of corresponding elements of C. oarri (Figs. I9 
and 20). Measurements of the type nuchal bones show no significant 
differences in proportions (Table 3). The long nuchal scute underlap 
of C. carri, which distinguishes this turtle from C. williaxnsi and C. 
concinna (Rose and Weaver, I966), is also evident in the type of C. 
caelata (Fig. I9, D and E) . Examination of more than 100 nuchal bones 
of C. oarri reveals that the length/width ratio of the nuchal scute 
underlap varies from 1.0 to 2.3- The ratio for the type of C. caelata 
is 1.0, within the limits of variation in the McGehee series. Sulci on 
the dorsal surface of the caelata nuchal are aberrant (Fig. I9A); these 
measurements are of little value. The moderate gular scute overlap of 
C. oarri, a character stressed by Rose and Weaver (I966) as highly 
diagnostic, is not significantly different from that on the type 
epiplastron of C. caelata (Fig. 20, A and C). The epiplastral lip is 
pronounced in both C. oarri and C. caelata (Fig. 20), although within 
the large McGehee series the shape and size of this lip are clearly 
variable. In both C. caelata and C. oarri the peripherals, and hence 
the carapacial margins, are un-notched. Furthermore, in both descriptions 
the authors characterize the posterior peripherals as horizontally 
flared (i.e., with a concave upper surface). Finally, all of the 



71 



Table 3- Maximum measurements (mm) of holotype 
nuchal bones of Chrysemys caelata and 
C. oarvi. 



caelata cavvi- 

length A6.6 hh.k 

width 53.3 52.0 



r'r'°' 23.3 23.0 

border 



72 



elements assigned to C. oaelata represent turtles within the size range 

of C. carri (largest plastron in McGehee series, 295 mm; estimated carapace 

1 ength, 338 mm) . 

I am unable to compare the remaining diagnostic characters of C. 
oarri--the slight median posterior keel and the notched rectangular pygal 
bone--with C. caelata as Hay's (1908) type series includes neither neurals 
nor pygal. Hay (1916) later erroneously assigned the neural (FGS \/-2599 = 
olf 3^25) of a Deirochelys to C. caelata; the impact of this error was 
discussed in Chapter II. 

There appear to be no significant morphological differences between 
C. oaelata and C. oavri. Furthermore, considering the geographic proximity 
of the type localities (3^ km) and the approximate strat igraphi c 
equivalence (Webb, \%k\ Hirschfeld and Webb, 1968) of the horizons, the 
names undoubtedly refer to the same species. Chvysemys carri Rose and 
Weaver is therefore a synonym of Chrysemys caelata (Hay). 

Rel at ionships 

Patterns of shell sculpture, scutellation and shell morphology of 
C. caelata are similar to those of C. nelsoni; the two species are surely 
closely related if not synonymous. Nevertheless, they may usually be 
distinguished by the shapes of the entoplastron and first suprapygal 
(Fig. 21). In addition differences in the ratios of maximum gular scute 
length/epiplastral lip width {oaelata x, 0.66; nelsoni x, 0.80) and 
posterior width of first suprapygal /greatest width of second suprapygal 
{oaelata x, 0.63; nelsoni x, Q.kk) are significant (Student's t-test, 
p < 0.025). Also, the pygal bone of C. caelata is more truncate, less 



73 



deeply notched, and usually less deeply overlapped by the fifth vertebral 
scute than that of C. netsoni. 

Rose and Weaver (I966) assigned two mandibles (UF II086 and UF IIO95) 
from McGehee Farm to C. oaelata. Based on its flattened verntral surface, 
which is nearly continuous with the scar marking the insertion on the 
dentary of the M. occ ip i to-squamoso-maxi 1 lar i s [(= Temporalis of 
Wiedemann, and M. temporalis of Bojanus, Stannius, Cuvier, and Owen) 
Hoffman, I89O], mandible UF 11095 is similar to that of C. nelsoni and 
hence may represent C. aaelata. There are no indications of the serrated 
cutting edge, bicuspid notching, or broad alveolar surface characteristic 
of jaws of C. nelsoni. However, mandible UF IIO86 is less flattened 
ventral ly, the muscle scar does not come close to reaching the ventral 
surface, and the entire jaw is foreshortened (hence, wider) relative to 
C. nelsoni; these conditions are similar to those of C. ooncinna (as well 
as C. floridana) and the jaw likely represents its predecessor, C. 
williamsi. For the present, as no skull material positively referable to 
C. oaelata is known, this species must be considered specifically distinct 
from C. nelsoni. 

Further study is also needed to clarify the relationship of C. 
oaelata with other Pliocene and Early Pleistocene Chrysemys, particularly 
C. idahoensis from Blancan beds in Idaho (Gilmore, I933; Rose and Weaver, 
1966) and C. hilUiof the Kansas Lower Pliocene. The latter has been 
associated by most authors with the Traohemys line (Cope, I878; Hay, 
1908; Adler, 1968; F. Rose, pers. comm.), and certain characters (e.g., 
notching of pyga 1 and peripheral bones) would seem to distinguish it 
from C. oaelata. 



Ih 



Discussion 

Chrysemys oaelata is now known from a large series of nearly complete 
shells from McGehee Farm in addition to the few original elements from 
Mixson's Bone Bed. The characteristics described by Rose and Weaver 
(1966) may be added to those listed by Hay (I908) to give us an improved 
definition of this species. Other than these two descriptions, C. caelata 
has been mentioned few times in the literature. Of the 12 additional 
Mixson elements Hay (1916) assigned to this species, at least the neural 
and probably the epiplastron represent De-iroahelys . Four of these, 
including the epiplastron (which, however, is pictured), must tentatively 
be considered lost. I have examined the remainder and, except for the 
neural, agree with Hay's assignment (although the peripheral he refers to 
as the second left is actually a first left). Gilmore's recognition of 
C. caelata from Melbourne, Florida (Hay, 1923, 1927), which Hay believed 
to be Aftonian (Late Blancan) but now generally considered Rancholabrean 
(Hibbard et al., I966; Webb, 197^), is extremely doubtful. I have seen 
only C. soripta from this site. All other references to C. caelata 
(Auffenberg, 1955, 1963; Coin and Auffenberg, I955) refer to a series 
of elements from Ha i 1 e Vl, another Florida Pliocene site only 3 i<m 
east of McGehee Farm. Many of these elements (a neural and the proximal 
ends of several pleurals) actually represent the genus Deivoohelys 
(Chapter 11). Allocation of the remaining fragments (mostly peripherals 
and distal ends of pleurals) is difficult; tentatively I consider at 
least some of them to represent C. caelata. Recent excavation of a 
newly discovered Pliocene site (Love Bone Bed) at Archer, Alachua County, 
Florida (29°33'N, 82°3rW; sec. 9, TllS, RI8E) has yielded large quantities 



75 



of turtle material, much of which can be definitely assigned to C. caelata; 
to date, two partial shells (UF20869 and 20870) as well as numerous 
unassociated elements representing this species have been removed. 

Ckrysermjs caelata, then, is a rugose Chrysemys presently known from 
four PI iocene sites ;encompassing a total area of only kl square kilometers 
in adjacent Alachua and Levy Counties, Florida. Recent workers (Auffenberg, 
1963; Hirschfeld and Webb, 1 968 ; Webb, I969 and pers. comm.) have 
assigned all four sites to the Alachua Formation. Although the "formation" 
is time-transgressive, including sediments ranging in age from early 
Miocene through Recent, fossils representing distinct faunas within it 
are often unmixed (Simpson, 1930; Cooke, 19^5; Vernon, I95I; Auffenberg, 
1963; Puri and Vernon, 1964; Hirschfeld and Webb, I968). I therefore 
conclude that Chrysemys caelata is indicative of eiarly Hemphill ian 
faunas (Hirschfeld and Webb, I968; Webb, I969) of the Alachua Formation. 
Morphologically, C. caelata appears to be ancestral to C. nelsoni, 
a relatively common turtle of both Pleistocene and Recent Florida. Like 
nelsoni, caelata was probably characteristic of still or slow-moving 
water. Likewise, caelata' s coeval at McGehee Farm, C. williamsi (Rose 
and Weaver, I966), appears, on the basis of shell morphology, to be the 
Pliocene forerunner of C. concinna, a stream- i nhab i tant (Crenshaw, 1955) 
of Pleistocene and Recent Florida. These two lines [floridana- concinna 
species group and rubriventris species group, including C. nelsoni) 
within the subgenus Pseudemys (sensu McDowell, 1964) were therefore 
distinct in north peninsular Florida by Middle Pliocene. Nevertheless, 
Crenshaw(l955, 1965) indicates that in rare instances in the southeastern 
United States reproductive isolation of the two groups has either 



76 



partially broken down or not yet fully developed. That C. caelata and 
C. williamsi were found in separate concentrations within the McGehee 
deposit (Rose and Weaver, I966) suggests that although sympatric they 
may have been predominantly allotopic (as are C. nelsoni and C. conainna 
today in this region). Sed imentology and associated faunas in the sa 
deposits (Webb. 1964; Hirschfeld and Webb. I968) support this hypoth 



me 



^ 


C> 


E 

o 


r— 


f- 


1_ 


(TJ 


i- 


1+- 


X- 


m 




o 


u. 


« 


13 




■+^ 


c 


0) 


« 




fl) 


v~^ 


U- 


-C 


W 


n 


0) 


f^ 




ts 


n 


tn 


o 





— - 


(3 (^ 






1-^ CO 




, — 


(» oo 




TO 


a o 


, — s 


!_ 


ly cvj 


<_) 


+J 




'•^ — ■ 


C 


to u. 




a) 


Si=> 


o 


> 


t= 


r~- 




0) "O 


CO 


-o 


CO c 


o 


c 


Si ID 


rsi 


03 


?H 





. 


O CNJ 


-o 


m 


oo 


a) 




<u CO 


m 


*. 


a. o 




< 


>-Cv4 


(U 


^ — ■ 


•M U- 


r 




Z2 


a 


,— 


(U 


en 


ro 


-c -^ 




i/i 


4-' ?s 


flj 


i_ 


?H 


> 


n 


u- « 


C) 


o 


o o 


-J 



78 



/''■ V 





V^' 




O v£) 




invo 




CM O 




o 




2: CM 




z 




to u. 




Z3 Z3 




G "^ 




4^ ?H 




« ?^ 




vo G 


- — ^ 


Q) <:j 





G 




• 


1 — 


to 




03 


X 


SiT3 


■ 


E c: 




ca nj 




CO 




Si— 


' — «. 


^ — 


f— 


rC 03 


(U 


to 1- 


1- 



TO •- •- 
.— in ui 

Q. 1- 1- 

Q) O O 

T3 -C 



80 






^^'-^i. 




Figure 21. Ventral surfaces of entoplastra (A, C) and dorsal 
surfaces of suprapygal bones (B, D) of Chrysemys 
nelsoni (A, B) and C. oaelata (C, D) . 



82 







B 







( 



D 



\ 



CHAPTER IV 

A REEXAMINATION OF THE CERYSEMYS SCRIPTA GROUP BASED ON 

FOSSIL EVIDENCE 



The systematic status and relationships of Chvysemys soripta and 
its close relatives have been a major center of controversy among turtle 
evolutionists. Most early workers (Agassiz, 1857; Gray, I87O; Hay, I908; 
Gilmore, 1933) assigned the C. scripta complex full generic standing as 
the genus Trachemys , although Boulenger (I889) included it in Chrysemys. 
Subsequent workers (Carr, 1952; Williams, 1956; Loveridge and Williams, 
1957) incorporated the group into the genus Pseudemys (as had Cope, I878) 
but made few statements about intrageneric relationships above the 
species level. It was not until 1964 when McDowell considered Pseudemys 
and Chrysemys {sensu striata) congeneric that subgenera were formally 
recognized. The subgenus Trachemys included all members of the C. scripta 
complex; the subgenus Pseudemys included two series of turtles--the 
floridana series (C. floridana and C. conainna) and the rubriventris 
series (C. rubriventris, C. alahamensis and C. nelsoni) --both more 
closely related to each other than either is to Trachemys; finally, the 
type species of the genus, C. picta, was considered a third distinct 
subgenus, Chrysemys. 

Weaver and Rose (1967) examined the ideas of McDowell (1964) using 
fossil and extant material and concluded that his subgenera were invalid. 
They developed a new phylogeny for the genus in which they cons idered the 
North American C. scripta group to be more closely related to the C. 
rubriventris complex, and the West Indian, Mexican, Central and South 

83 



8^ 



American turtles previously considered races of C. scripta to be more 
closely related to C. ftoridana and C. concinna. Their failure to 
recognize the importance of convergence of specific adaptations, as well 
as the false premises on which they delineated "ancestral characters," 
leave their conclusions suspect. 

One of the major drawbacks to a vertical study of the C. scripta 
complex has been the absence of skulls of extinct forms; in fact, only 
two fossil emydid skulls from all of North America have been previously 
reported (Hay, 1 908 ; Gilmore, 1933), and the identities of these are 
ambiguous. McDowell's (1964) conclusions, drawn almost entirely from 
cranial characters, have therefore been nearly impossible to apply to 
fossil forms, the taxonomy of which has necessarily been based almost 
exclusively on shell osteology. In this paper I report previously 
unrecognized skull material from the Florida Pliocene which casts new 
light on some of these problems. Furthermore, only very limited attempts 
have been made to relate fossils from Florida to those from the Great 
Plains; this must be done if pal eontolog ical species are to have 
biological meaning. My purpose, therefore, is not to revise the genus 
again, but to analyze all available fossils to point out problems with 
some of the earlier schemes, and to reinterpret intrageneric relationships 
accordingly. Of primary interest here is the relationship of the C. 
scripta complex to other members of the genus. The term "North American 
C. scri-pta,*' as used in this paper, includes only those turtles occurring 
in the continental United States. All fossils, unless otherwise noted, 
are in the vertebrate paleontology collection of the Florida State 
Museum (UF) . 



85 



Tvachemiis In the Pleistocene 

Probably due to their abundance in the Pleistocene of Florida and 
Texas and the ease with which they may be recognized, fossils of the C. 
scripta group (i.e., Traohemys) have been known longer and studied more 
extensively than those of most other southeastern emydid turtles. Hay 
(1908, I9I6) recognized eight extinct species from Pleistocene deposits 
in Florida and Texas that he assigned to this group: Traohemys euglypha 
(Leidy), T. saulpta Hay I908, T. ? jarmani Hay I9O8, T. petrolei 
(Leidy), T. bisomata (Cope), T. tvulla Hay I908, T. ? delioata Hay 
1916 and T. ? nuohooarinata Hay I9I6. Weaver and Robertson (I967) 
correctly placed six of these names in synonymy with C. saripta and 
incorporated them, as well as other Florida Rancholabrean material, in 
their new combination C. s. petrolei. The remaining two names represent 

fossils incorrectly assigned to Traohemys: T. nuohooarinata = Terrapene 

Carolina (Auffenberg, 1958) and Traohemys jarmani = C. nelsoni (Chapter 
II). The only other Traohemys recognized by Hay was T. hillii (Cope) 

from the Pliocene Loup Fork beds of Kansas; it is discussed further 

below. 

Because of intraspeci f ic variation within Recent, Rancholabrean, 
and irvingtonian C. soripta, the utility of C. s. petrolei as a reliable 
stratigraphic tool in Florida is meager. Justification for giving the 
Rancholabrean fossils separate taxonomic status must be questioned. 
Weaver and Robertson (I967) distinguish C. s. petrolei from all other 
C. soripta by only two characters: its larger average size and greater 
carapacial rugosity. They admit, however, that these are "minor 
distinctions." They also state that "the extensive rugosity and 



sculpturing of the RanchoJabrean fossils is often present in large, extant 
specimens of C. s. saripta." Furthermore, they add that "an additional 
series of fossils from Ichatuckenee Springs" shows "a size gradation from 
typically large Rancholabrean nuchals to smaller ones which, in the absence 
of mineralization, are indistinguishable from those of extant C. s. 
scripta" (italics mine). 

Neither of the two characters used to diagnose C. s. petrolei is 
reliable. Although the shells of Rancholabrean C. saripta are often 
more rugose than their modern counterparts, the character is subject to 
extreme variability in both temporal groups. It is not uncommon to find, 
in peninsular Florida, living C. saripta whose shells are more rugose and 
more deeply insculpted than those of many fossil C. saripta. 

The second and chief character by which Weaver and Robertson (196?) 
define C. s. petrolei- \arger average size than extant C. saHptaseems 
to me insufficient to serve alone as the basis for a separate taxon. 
The larger average size of turtles in the Pleistocene versus those today 
is not unique to C. saripta. It can be documented not only in other 
species of Chrysemys in Florida but also in most other genera; e.g., 
Deiroahelys (Chapter II). Graptemys (Chapter I), and Terrapene (Auffenberg. 
1958). Many reasons can be speculated to account for larger size in the 
Pleistocene: climatic differences as they affect heat loss and retention 
by poikilotherms; size-selective predation by man or other predators; and, 
nutritional differences in diet, etc. Body size is a complex phenomenon 
and should not be used as the sole criterion for establishing additional 
taxa. 

The designation of a temporal subspecies must be made with extreme 
caution (see Mayr, I969). Although the concepts of temporal and geographic 



87 



subspecies are not necessarily equivalent, they must nonetheless be 
compatible when applied to one species. The designation of a temporal 
subspecies that might in itself encompass more than one geographic 
subspecies is more apt to cause confusion than to increase understanding. 
This is precisely the case with C. s. petrolei, as pointed out previously 
by Preston (1971). Two distinct subspecies occupy the purported range 
of C. s. petrolei (Florida to Atascosa County, Texas) today: C. s. 
elegans in the west and C. s. saripta in the east. The possibility 
certainly exists that more than one race of C. saripta occupied this 
region during late Pleistocene, so that C. s. petrolei in Texas may have 
been subspeci f ical ly distinct from C. s. petrolei in Florida. Essentially 
the same problem arose when Preston (I966, 1971) recognized C. s. 
hisornata (by which he meant C. s. elegans-] ike turtles from the 
Irvingtonian mammalian age) from both Florida and Texas. 

It is unfortunate that the diagnoses of the modern subspecies of 
C. scripta rely almost entirely upon color pattern, as they thus cannot 
be compared directly with the fossils. The osteological characters used 
by Preston (1966) to distinguish these forms have some val ue--part icular ly 
the development of the middorsal i<eel and the relief of the nuchal 
lamina--though i ntrasubspec i f ic variation is too great to consider them 
infal i bl e. 

Interestingly, C. scripta from Irvingtonian deposits in Florida 
and Texas appear to differ osteolog ical ly in the same way that the 
subspecies C. s. saripta and C. s. elegans differ in these regions today. 
Thus, a nearly complete shell (MUVP 45^6) from the early Irvingtonian 
Seymour Formation, Burnette Ranch, Knox County, Texas, lacks a pronounced 



middorsal keel and has a relatively flat nuchal scute and low profile. 
A partial shell (UF 2l802) from Coleman MA, Sumter County, and a series 
of elements from Haile XV I , Alachu£< County--both Irvingtonian deposits in 
Flor ida--d i spl ay the raised nuchal scute, pronounced middorsal keel, and 
slightly higher profile characteristic of C. s. scripta. It appears, 
therefore, that these two modern subspecies became established no later 
than early Pleistocene. Any scheme that assigns these fossils subspecific 
standing distinct from modern forms seems to me biologically unrealistic 
and an artifact of choosing one's own lifetime as a significant point of 
reference in geologic time. 

I suggest, as implied by Preston (1971), that until we achieve a 
better understanding of the C. saripta group (possibly through the 
discovery of more fossils), the recognition of fossil subspecies is 
unwarranted and can only lead to further confusion. Fossils resembling 
C. s. elegans or C. s. scripta should be referred to the species and their 
apparent affinities with modern subspecies discussed, but assignment of 
temporal subspecific names is presently premature. 

Pleistocene fossils of C. saripta provide us with information on 
the species' former distribution and the occurrence of subspecies. 
However, for knowledge of the evolution of Trachemys we must look back 
to at least the Pliocene. 

Trachemys in the Upper Pliocene 

Systemat ics 

Weaver and Robertson (1967) described Chrysemys platymarginata as 
a member of the Trachemys group from Haile XVA, Alachua County, Florida. 



Although the fauna of this site was then thought to be Irvingtonian 
(Pleistocene), Robertson (1976) has since reinterpreted it as Blancan 
(latest Pliocene). Other peninsular Florida sites of Blancan age also 
contain C. pZatymarginata (Weaver and Robertson, 1967). Based on shell 
morphology (doubly- toothed peripheral bones, extensive gular scute 
overlap and nuchal scute underlap, highly sculptured nuchal bone, and 
well developed median keel on the carapace), Weaver and Robertson 
properly assigned C. pZatymarginata to the C. scripta complex (subgenus 
Traohemys) . They believed no skull material of C. pZatymarginata was 
available. However, a small box of cranial fragments from Haile XVA, 
collected in 1964 by S. D. Webb, J. Robertson, and R. Allen, contains 
parts of the skulls of this species, and a chelydrid and trionychid. 
The fragmentary materia] was reassembled and compared to literature 
descriptions (Hay, I908; Gilmore, 1933; McDowell, 1964; Weaver and 
Robertson, I967) and Recent specimens. Much of the cranial anatomy of 
the fossil emydid can be observed from three partial skulls (UF 21888, 
21892, and 21963); UF 21888 includes both dentaries as well. Unassociated 
fragments (UF23920 and 23921) from several other skulls are also 
available. Since other emydids (e.g., Chrysemys ooncinna and DeivooheZys) 
are also present in the Haile XVA fauna, the emydid skull fragments must 
be examined closely before specific assignment is made. 

McDowell (1964) diagnoses the genus Chrysemys by the following 
cranial characters: "triturating surface of maxilla with sharply defined 
middle ridge; anterior edge of inferior process of parietal thin; 
posterior end of pterygoid usually not in contact with exocc i p i tal ." 
Examination of the Haile XVA skulls confirms their identity as Chrysemys 
as defined by McDowell (1964). 



90 



Additionally, according to the criteria set forth by McDowell (196^.) 
and modified by Weaver and Rose (I967). the skulls are clearly associated 
with the Tvaahenys line within the genus. McDowell characterizes the 
subgenus as follows: ventral surfaces of dentary relatively rounded, 
middle ridge of upper triturating surface lacking an anterior cusp, 
maxilla not sutured to quadratojugal (= McDowell's squamosal), pterygoid 
extending back near the exoccipital, and crista praetemporal is heavier 
than in other Chrysemys. The Ha i 1 e XVA skulls are clearly assignable 
to the TTaohemys line within the genus Chvysemys and may thus be confidently 
referred to C. platymarginata , the shell of which also reflects its 
Trachemys affinities. Two other cranial characters that I find consistent 
with extant Trachemys are the recessed vomer, which does not participate 
in the alveolar surface (unlike the C. rubriventris complex), and the 
broad pterygoids (O.3I to 0.32 times skull width, as compared to 0.23 to 
0.28 for species of the C. rubriventris group and 0.32 to O.38 for C. 
soripta; Fig. 22A. A symphysial ridge on the mandibular alveolar surface 

appears to be consistently absent or inconspicuous in all living and 
fossil Trachemys. Finally, the posterolateral region of the dentary of 

C. platymarginata, as in C. scripta, rises abruptly to meet a relatively 

higher coronoid bone than is present in either the C. rubriventris or C. 

floridana series. 

In the above, as well as most other characters, the skull of C. 

platymarginata differs little from that of its presumed descendant, C. 

scripta. Certain differences, presumably adaptive, do exist, however. 

Most obvious are the much broader maxillary surfaces of C. platymarginata 

(Fig. 22B); this is principally due to expansion of the posterior part ., 



91 



of the alveolar surface medial to the longitudinal maxillary ridge. 
Alveolar expansion is present in all Ha i 1 e XVA maxillae but seems most 
highly developed in the largest individuals. The alveolar surface 
incorporates a lateral projection of the palatine bone as it does in 
extant members of the scripta, floridana and rubriventris groups. The. 
supraoccipital of C. platymarginata appears slightly larger than in Recent 
C. scripta, though C. platymavginata cranial material is insufficient to 
test this statistically. This feature, coupled with greater lateral 
expansion of the parietal roof in C. platymarginata, would provide 
additional surface area for attachment of the adductor mandibulae 
externus musculature (Gaffney, 1972). 

Most of these morphological features indicate adaptation to coarser 
or harder foods than those eaten by the omnivourous C. scripta. In 
particular, they suggest a turtle with feeding habits similar to modern 
C. nelsoni (unpublished data show post-hatchl ing C. nelsoni to be entirely 
herbivorous). The broad pterygoids indicate that animal food may still 
make up a significant part of the diet, however. 

With knowledge of the skull of Chrysemys platymarginata, its 
relationships to other species in the genus can now be studied. Other 
than Rogers' (1976) report of C. scripta from Texas, the only Blancan 
turtle known which may be a member of the C. scripta group is Chrysemys 
idahoensis. Described by Gilmore (1933) from the Hagerman lake beds 
(Upper Pliocene) of Idaho, the species is based not only on an entire 
(though partially crushed) shell, but also on a remarkably preserved 
and excellently illustrated skull (USNM 12059). Although unsure of its 
generic identity (whether Pseudemys, Trachemys, or Graptemys) , Gilmore 



92 



believed the skull, with its broad alveolar surfaces, bore closest 
resemblance to C. vuhriventris. He pointed out, however, that the skull 
differs from C. vuhvlventvis by its broader pterygoids and less pronounced 
median alveolar ridge with finer dent iculat ions . These features are 
diagnostic of the Tvaoherrys group. Hay's (I908) improper description 
of the lower jaw of that group led Gilmore to say that C. idahoensis could 
not belong to Traohemys. Rose and Weaver (I967) noted the strongly 
notched peripheral bones of C. idahoensis, but nonetheless concurred with 
Gilmore's opinion of its affinities with the C. vuhriventris series. 
Zug (1969) correctly points out that the cranial characteristics and 
geographic distribution of C. idahoensis are actually more similar to 
C. saripta than to the C. rubriventris line. The primary feature by 
which previous authors associated C. idahoensis with the C. vuhviventvis 
lineage is the broad alveolar surfaces of their jaws. Broad alveolar 
surfaces have occurred in the C. scviyta lineage in the past, however, 
as noted above in C. platymarginata . This characteristic is an adaptive 
trophic feature which should not be employed per se to distinguish the 
two groups. Additionally, in all Recent species assigned to the C. 
ruhviventvis group (c. rubriventris, C. nelsoni, and C. alabamensis) , 
the vomer forms an integral part of the alveolar surface; the vomer of 
C. idahoensis, like that of C. scripta, is recessed. All other features 
used earlier to assign C. platymarginata to the Traohemys group are also 
present in C. idahoensis. Gilmore (1933) points out the occurrence of 
carapacial rugosity in the paratype of C. idahoensis and the wide pterygoids 
of the type skull, both characteristic of C. scripta. The posterior 
tapering of the shell and elevated (tent-like) pygal bone of the holotype, 



93 



considered as a species characteristic by Gilmore (1933) are probably 
artifacts of post-mortem compression; both features are more nearly 
"normal" in the paratype. 

Although Zug (1959) contends that the broad alveolar surfaces, 
coupled with robust inferior parietal processes, indicate mol 1 usci vory , 
I disagree. For reasons stated earlier i believe that these characteristics 
in the genus Chrysemys are adaptations for herbivory. 

The type skull figured by Gilmore (1933) appears at first glance 
extraordinarily large for Chrysemys and particularly for Trachemys. 
Comparison with Recent C. nelsoni and extrapolations for Recent C. 
soripta of equivalent size (C. saripta does not reach this size today) 
reveals that the skull of C. idahoensls is not significantly longer than 
that of C. scripta or C. nelsoni though it is proportionately wider. 
The skull of C. idahoensis rises abruptly posterior to the f ronto-par ietal 
suture, effectively creating a higher and deeper supraocc i pi tal crest. 
Whether this is an artifact of illustration or preservation, an aberrancy, 
or a real adaptation allowing for attachment of hypertrophied jaw 
musculature, remains to be determined upon collection of additional 
cranial material . 

I conclude that C. idahoensis is a valid member of the C. scripta 
group; it is known from the Upper Pliocene of south-central Idaho, by 
Gilmore's (1933) original material and by six additional elements 
described by Zug (1969). 

Two extinct species, C. idahoensis and C. platymarginata , both 
clearly represent the Trachemys group. Both occur in Upper Pliocene 
(Blancan) deposits, but their type localities are 33^5 km apart. The 



9^ 



question arises as to whether these two turtles represent contemporaneous 
but distinct lineages within the C. saripta group, or whether they might 
actually represent two populations of a single widespread species. Were 
it to be based only on records from Idaho and Florida, the latter 
hypothesis could be seriously questioned on geographic grounds alone. 
However, Zug (1969) identified the anterior half of a plastron (UMMP 
\J-k2603) from the late Hemphill ian Wolf Canyon area of Meade County, 
Kansas as C. iddhoensis. Additionally, among material collected by 
J. A. Holman at Devil's Nest Airstrip, Knox County, Nebraska, are an 
eleventh left peripheral bone (MSU VPS^Z) and a small xiph i plastron 
(MSU VP832) which represent Traahemys and are indistinguishable from 
C. idahoensis. The material from this site is tentatively believed to 
be Hemphillian (J. A. Holman, pers. comm.). Finally, a series of 
approximately 30 elements (MUVP 9267), including four nuchal bones, 
from the Upper Pliocene (Blancan) Beck Ranch local fauna in Scurry 
County, Texas, were assigned by Rogers (1976) to C. saripta. These fossils 
are here reassigned to C. iddhoensis on morphological (e.g., peripheral 
bone rugosity and flare) and temporal grounds. The Kansas and Nebraska 
localities fall approximately midway between the type localities of C. 
idahoensis and C. platymarginata , roughly 1^00 km from the former. its 
additional occurrence in Texas suggests C. idahoensis must have occupied 
much of midcont i nental North America during the Pliocene. 

Today only a single polytypic species of Traahemys, Chrysemys saripta, 
occurs along the Atlantic Coast from Virginia to northern Florida, westward 
to southern Texas, and northward to northern Kansas, Missouri, and 
Illinois (Conant, 1975). Thus, it is not unusual for species of this 



95 



group to be distributed over extensive areas and wide latitudinal ranges. 
From a geographic standpoint, therefore, it would not be surprising if 
C. platymavginata and C. idahoensis represent a single widespread late 
PI iocene species. 

Morphologically C. idahoensis and C. platymarginata are similar. 
Both are large turtles (max CL greater than 300 mm) with a faintly rugose 
carapace, notched peripheral bones, incised nuchal scute, and extensive 
plastral scute overlap. The skulls of both are like C. saripta but with 
a more extensive parietal roof, larger supraocc ipi ta 1 , and broader 
alveolar surfaces. The distinct double-notching of the posterior 
peripheral bones characteristic of C. savipta is evident in C. platymarginata; 
although less pronounced, the condition does occur in some peripheral 
bones of C. idahoensis (Zug, 1969; Rogers, 1976). Likewise, the mid- 
dorsal keel and carapacial sculpturing are more pronounced in C. 
platymarginata than C. idahoensis. However, these differences are no 
greater than those existing between the Recent subspecies of C. 
scripta—C. s. elegans and C. s. scripta. Where such minor differences 
do occur, C. idahoensis more closely resembles the northern and western 
C. s. elegans, and C. platymarginata the southeastern C. s. scripta. 
Should C. idahoensis and C. platymarginata actually represent a single 
species, the evolution of what we recognize as subspecific differences 
may have begun by at least the late Pliocene. 

I suggest that Chrysemys idahoensis (Gilmore) and Chrysemys 
platymarginata Weaver and Robertson represent a single widespread species 
of the C. soripta group. This is supported by their morphological 
similarity, approximate contemporaneity, and existence of geographically 



96 



intermediate populations. Chrysemys platymarginata should therefore be 
placed in the synonymy of C. iddhoensis. Based on present subspecies 
distributions and minor morphologic differences in the Pliocene, it is 
likely that populations from Idaho and the Great Plains were subspec i f ical ly 
distinct from those in Florida. 

Based on minor cranial and carapacial differences, Chrysemys 
idahoensis is considered distinct from C. scripta (see Weaver and 
Robertson, 1967, for carapacial differences). The carapace of C. 
idahoensis is generally less rugose and more posteriorly flared, and the 
upper triturating surface and parietal roof are more extensive than 
those of C. scripta. These and other minor differences may be chronoclinal, 
and the point in time at which we separate the species must necessarily be 
in part an arbitrary decision. An examination of Tvaahemys fossils from 
five Irvingtonian deposits (Coleman ilA, Haile XVI, Inglis lA, Pool 
Branch II, and Punta Gorda) and six Blancan deposits (Haile lA and 
XVA, Port Charlotte, and Santa Fe River I, IB, and VIM) in peninsular 
Florida, and one early Irvingtonian site (Burnette Ranch, Seymour 
Formation, Knox County) in Texas, reveals almost exclusively C. scripta 
phenotypes in the Irvingtonian and predominantly C. idahoensis phenotypes 
in the Blancan; some phenotypic mixture as well as i ntegradat ion occurs 
in both. The shift from C. idahoensis to C. scripta is therefore believed 
to have occurred from late Blancan to very early Irvingtonian. 

Pal eoecology 

Chrysemys idahoensis was a large, mostly herbivorous, polytypic 
species that represented the C. scripta group over much of North America 



97 



during the Upper Pliocene. Its great latitudinal range may suggest a 
more homogeneous climate in North America at this time, although C. 
saripta today is adapted to a wide variety of climatic regimes. The 
presence in the Hagerman fauna of mammals with extant southern 
distributions likewise indicates that winters must have been milder 
than at present (Zakrzewski, I969). Fossils of otters (Bjork, 1970), 
beavers and voles (Zakrzewski, I969), large numbers of fish (Miller 
and Smith, I967), frogs, water snakes (Holman, I968), water and shore 
birds (Brodkorb, 1958; Feduccia, 1967; Murray, 1967), as well as the 
pollen record (Leopold in Weber, I965), suggest a slightly wetter climate 
and greater abundance of fresh water in the Hagerman area during deposition 
than at present (Zakrzewski, I969). Chantel 1 (1970) concludes from the 
frog fauna that Hagerman was probably a warm, wel 1 -vegetated floodplain 
habitat. In light of this evidence, the occurrence of a C. saripta-] ike 
turtle in Idaho is not surprising. 

Pleistocene glaciation must have exterminated the northernmost 
populations of C. idahoensis. Failure of the species (or of its presumed 
descendant C. saripta) to reestablish northern populations may have been 
due 1;o establishment of lower (harsher winter) temperatures coupled with 
development of more xeric condition?. Drainage changes that must have 
occurred (Miller, 1965; Taylor, I966) would have had profound effects on 
the distributions of all aquatic vertebrate populations. Reestabl ishment 
may have been further hindered by the successful invasion of another 
aquatic emydid turtle, C. piata, although C. piata and C. saripta coexist 
extensively today. 

Populations of C. saripta in the Florida peninsula (as far south 
as Palm Beach County) also disappeared later in the Pleistocene. Here, 



competition with several, more efficient large herbivorous congeners (C. 
floridana, C. nelsoni, C. conoinna) may have been the determining factor. 
Alveolar surface width reduction and related changes between C. 
idahoensis and C. scripta may indicate a shift from more specialized 
herbivory to generalized omnivory in response to the immigration or 
evolution of superior competitors. The shift towards increased carnivory 
may have necessitated the reduction in body size which has occurred in the 
C. sar-ipta line during late Pleistocene and Recent times. Should this be 
the evolutionary strategy that led to modern C. saripta, the species' 
abundance over most of its large range today attests to its marked success. 

Systematic Conclusions 

McDowell's (1964) characters defining the subgenus Trachemys are, 
as pointed out by Weaver and Rose (I967), either invalid or insufficient, 
although some (dentary shape, anterior triturating cusp and maxillary- 
quadratojugal junction) are helpful when used in conjunction with other 
characters. However, rather than redefine McDowell's subgenera, Weaver 
and Rose (1967) state that his subgeneric groupings are totally invalid. 
They contend that North American C. saripta (including fossil forms) are 
actually more closely related to the C. rubriventris series than either 
is to the C. floridana series. I cannot agree with their interpretation 
for several reasons. 

Firstly, they accepted the opinion of most previous authors (Hay, 
1908) that the Eocene emydid fossil turtles from the Bridger and Uinta 
Formations, assigned to the genus Echmatemys, were ancestral to most 
North American emydid genera, including Chrysemys. Characters of the 



99 



skull and shell of Eohnatemys were accordingly considered to represent 
ancestral conditions where they occurred in Chrysemys spp. Unfortunately, 
their basic premise is probably incorrect. Many of the shells, and also 
the only known skull, of Eahmatemys appear to represent the Neotropical 
emydid genus Rhinoalemys (= Callopsis of Smith et al., 1976), considered 
by McDowell (196A) to be the only New World genus in the subfamily 
Batagurinae. Characters of "Eahmatemys" are thus of little value in 
determining relationships within the genus Chrysemys. 

The second problem with the interpretation of Weaver and Rose (1967), 
as well as of McDowel 1 (196't) in some instances, is the failure to 
recognize morphological convergence due to ecological adaptation. Such 
characters as shell shape, alveolar surface width, and scute overlap and 
underlap reflect specific adaptations to the environment and not necessarily 
phylogenetic relationships. In aquatic turtles, for example, narrow 
scute overlap and underlap, by decreasing turbulence of a submerged turtle, 
are adaptations to lotic environments (as witness their occurrence in 
such distantly related riverine turtles as Gvaptemys, Podoonemis, Chrysemys 
ooncinna, Trionyx, etc.). The fact that C. saripta and C. nelsoni have 
greater scute overlaps does not imply a close phylogenetic relationship 
but rather reflects evolutionary responses to similar ecologies (occupation 
of more lentic habitats). Alveolar surface width has already been shown 
to vary between closely related forms (C. saripta and C. idahoensis) and 
must therefore be treated very cautiously as a phylogenetic character. 
Such characters as strong peripheral bone notching have surely arisen 
independently in C. saripta and C. concinna and can therefore not be 
regarded as primitive to the entire genus. 



100 



Finally, as I have shown previously (Chapter III), fossil evidence 
does not indicate a close relationship of C. saripta to the C. rubriventris 
series as stated by Rose and Weaver (196?) and Weaver and Rose (196?). 
Furthermore, structure of the choanae imply a distinct dichotomy between 
the C. saripta series and all members of the C. flor-idana and C. 
rubriventris series (Parsons, I960). 

For these reasons the phylogeny proposed by Weaver and Rose (1967) 
should be rejected. McDowell's (1964) concept of the genus seems more 
realistic. Whether his subgenera would be better considered distinct 
genera must await study of the fossil record of C. piota. 

With knowledge of the Upper Pliocene Traohemys described in this 
paper and that of Gilmore (1933), I present the first diagnosis of the 
group that incorporates characters of both the shell and skull as they 
have evolved through time. This diagnosis does not attempt to include 
Mexican, Antillean, and Central and South American forms as their relation 
to the C. scripta complex remains unclear (Weaver and Rose, 1967). 

Characters clearly associated with all known North American 
Traohemys are pterygoids broad and extending posteriorly to exocci pi tal s; 
alveolar maxillary ridge without cusps or strong serrations; vomer not 
participating in alveolar surface; mandibular symphysial ridge absent 
or inconspicuous; anterior and posterior peripheral bones notched (some 
doubly); nuchal scute sulci incised deeply on nuchal bone; medial 
longitudinal keel present at least posteriorly; carapace but not plastron 
usually with some longitudinal wrinkles; gular and femoral scutes broadly 
overlapping plastron; and nuchal scute underlap longer than wide. The 
number of phalanges is unknown in fossil forms. 



101 



Further Problems 

Ciwysemys idahoensis is clearly not the progenitor of the Trachemys 
line; it appears too late in the fossil record and already shows most of 
the diagnostic characteristics, such as the doubly toothed peripherals, 
of modern C. soripta. Chrysemys hillii (Cope), as expanded by Adler 
(1968) to include C. lirrmadytes (Gal breath, 19^8), is known from the 
early Pliocene of Oklahoma and Kansas. Adler (I968) points out a number 
of similarities of C. hillii to C. soripta, including slight notching of 
the posterior peripheral bones; he speculates that the fossil species 
may be ancestral or closely related to C. soripta. Cope (I878) himself 
suggested such a relationship and was followed by Hay (1902, I9O8), who 
referred the species to Trachemys. I have not yet seen the types of 
either C. hillii or C. limondytes; however, from Hay's (I908) excellent 
photograph and Galbreath's (1948) illustration, I am inclined to agree 
withAdler's (I968) hypothesis. 

The presumably middle Pliocene Chrysemys inflata (Weaver and 
Robertson, 1967), still known only from peninsular Florida, remains 
ambiguous. Until further material is discovered I will fol low Weaver 
and Robertson's (I967) suggestion that "it was a specialized or aberrant 
species characterized by an extreme development of Traohemys features 
and not representative of the main evolutionary sequence leading to recent 
C. soripta." (p. -65) I additionally hypothesize that C. inflata represents 
a pre-Blancan isolate from C. idahoensis stock that, by genetic drift or 
perhaps in response to unique environmental stresses, developed its 
massive, gothic shell. its failure to persist, as well as its subsequent 
replacement in the peninsula during the Pleistocene by more conventional 



102 



C. saripta stock, may indicate overspec ial izat ion or ma ladapt i veness to 
a changing environment. It is not surprising that C. inflata is known 
only from peninsular Florida, where C. idahoensis (platymarginata) 
appears to have developed the largest and most massive shell within its 
range. I predict that C. inflata will not be found elsewhere. 

The recent discovery of James L. Dobie (pers. comm.) of the nearly 
complete carapace of a Chrysemys saripta (TMM 31081-280) from Bee County, 
Texas, in the collections of the Texas Memorial Museum, is confusing. 
The shell is reportedly from the Goliad Formation (late Miocene-early 
Pliocene). It is, however, remarkably similar to Middle and Upper 
Pleistocene C. saripta, particularly W\IPhS^(> (Texas I rv i ngton ian) . 
I suspect that, through a data mix-up or deposltional quirk, the shell 
may be Pleistocene. The alternative is that C. saripta existed as early 
as Middle Pliocene (perhaps including C. hillii) , and that turtles 
assigned to C. inflata and C. idahoensis represent a now extinct second 
line of Traohemys in North America. 

Finally, the status and relationships of the several Mexican, 
Central and South American, and Antillian extant forms assigned to the 
C. saripta complex (Williams, 1956; McDowell, 1964; Weaver and Rose, 
1967; Moll and Legler, 1971) remain to be determined. As shown 
previously many of the relationships suggested by Weaver and Rose 
(1967) are based on inadequate characters; hence, their conclusions 
regarding the relationships of the Neotropical turtles in this complex 
must be reexamined. Unfortunately, these turtles are essentially 
unknown as fossils. I suggest that they be retained provisionally as 
members of the C. saripta complex as put forth by Williams (1956). 



Figure 22A. Posterior palatal surface of_si<ull (UF 21963) of 
Chrysemys idahoensis from Halle XV. 



Figure 22B. Alveolar surface of right maxilla (UF 23921) of 
C. idahoensis from Ha i 1 e XV. 



\0h 





CHAPTER V 
PRE-PLIOCENE RECORDS OF THE GENUS CHRYSEMYS 

Emydid turtle fossils have been reported from only two of the 
several Miocene deposits known from Florida (Olsen. 1 964b) --Thomas Farm 
in Gilchrist County (Williams, 1953) and the Seaboard Air Line Railroad 
Company in Leon County (Olsen, 1964a). The fossils fron both sites were 
referred to the genus Chrysernys (sensu McDowell, 1964) and, in conjunction 
with its presence in an 01 igocene deposit (Patton, I969), mark the 
earliest known occurrence of the genus. In light of its wide distribution 
and significant role in the North American turtle fauna, these early 
records of the genus Chvysemys are of considerable importance. After 
examining fossils from these sites I find that a reappraisal of the 
material is necessary. 

Materials and Methods 

I examined all previously described Thomas Farm fossils from the 
Museum of Comparative Zoology (MCZ) as well as additional undescribed 
material from the Florida State Museum (UF) . The Seaboard fossils, 
formerly housed by the Florida Geological Survey, are now also in the 
Florida State Museum, as are the 1-75 fossils. Comparative skeletal 
material was examined from the Florida State Museum (UF) herpetology 
collection as well as the author's (DRJ) personal collection (the latter 
to be incorporated into the UF collection). 

105 



106 

Site Records for Chrysemys 

1-75 

Patton (1969) lists the occurrence of "Pseudemys sp." in this site, 
the only known nonmarine 01 igocene deposit in Florida. Unfortunately, 
some of the vertebrate fossils from this site have been lost (T. Patton, 
pers. comm.). Of the chelonian fossils available (UF I6897- I6899) , the 
only identifiable fragments represent land tortoises. 

Seaboard Air Line Railroad Company, Switchyard B 

Olsen (I96'*a) recognizes "pieces of the plastron and carapace of 
the terrapin Pseudemys sp. idet." from this Lower Miocene deposit in the 
Florida panhandle. Having examined all of the turtle material from this 
site, I find no fossils that can be assigned to any emydid genus. Several 
fragments labeled "Pseudemys hypoplast ron" proved, upon reconstruction, to 
be a large posterior peripheral bone of a sea turtle (UF 21952). The only 
other fossils labeled Pseudemys are actually fragments of a xiph iplast ron 
(UF 21953) and pleural bone (UF 21954) from two small tortoises 
(Testudinidae) . As Olsen (1964) reports the presence of the tortoise 
Geoohelone and four genera of marine fishes, neither reassignment contradicts 
his conclusions about the stratigraphy and paleoecology of the site. 

Having eliminated the 01 igocene and one of the two Miocene reports 
of Chrysemys , we must now examine the remaining Miocene record more 
closely. 

Thomas Farm 

Williams (1953) first noted the presence of emydid turtle fossils 
in the Thomas Farm Lower Miocene fauna and tentatively assigned them to 



107 

the Chrysemys floridana group. Rose and Weaver (I966) affirmed the generi 
designation but reassigned the fossils to the C. rubriventris group. 
However, none of these workers realized the antiquity of the fossil record 
of the genus Beirochelys (Chapter II), nor did they have any reason to 
suspect its occurrence in the Thomas Farm fauna. Confirmation of the 
presence of Deivochelys in this fauna (Chapter II) therefore necessitates 
a reexamination of all emydid material from this site. 

The inclusion of Chrysemys as a member of the Thomas Farm fauna, 
based only upon the original material (MCZ 3^32), would be questionable 
were it not for two adjacent neural bones (UF 21933) collected since 
Williams' paper. The two neural bones (Fig. 23), though possessing an 
anomalous common sutural border dorsal ly, are clearly the second and 
third neural bones of a relatively small Chrysemys (estimated carapace 
length 220). They are longer than wide (length:width ratios of 1.2 and 
1.1, respectively), unlike Deiroohelys (Chapter II), and possess no 
middorsal keel. Their dorsal surfaces are finely rugose, and the neural 
spines are low and robust. A posterior neural bone (UF 21935) almost 
certainly representing Chrysemys further indicates the absence of a 
strong keel. The lack of a keel or of any gross surface sculpturing is 
characteristic of McDowell's (I96i») subgenus Pseudemys and the C. omata 
group of the subgenus Traahemys. 

With the occurrence of Chj'ysemys in the Thomas Farm Miocene thus 
reaffirmed, we may suspect that most of the emydid fossils probably do 
represent this genus. However, until complete material represents both 
genera in the Miocene, it will be difficult to assign many of the 
individual Thomas Farm bones to either genus with certainty. Neverthe- 
less, as all of the unworm peripheral and pleural bones from the site 



exhibit a finely to moderately rugose surface, we may conclude with some 
assurance that the Thomas Farm Clwysemys possessed a rugose carapace 
similar to that of C. nelsoni and the Pliocene C. oaelata (Chapter III). 

Because of the importance placed on the nuchal bone by most 
chelonian paleontologists, I am obligated to comment on the Thomas Farm 
nuchal s. Without knowledge of the presence of Deirochelys in Thomas 
Farm, both Williams (1953) and Rose and Weaver (I966) naturally assumed 
them to represent Chrysemys. Rose and Weaver pointed to the resemblance 
of the nuchal scute underlap of the complete nuchal bone (see their Fig. 
2B) to that of C. nelsoni and the Pliocene C. oaelata (their C. oai>ri; 
see Chapter II). The similarity also holds for the two fragmentary nuchals, 
but, more importantly, equivalent nuchal scute underlaps for all three 
may be found in Deirochelys as well (e.g., UF Ukk; DRJ 300, 305). 
Furthermore, Rose and Weaver (I966) overlooked the importance of the 
laterally expanded first vertebral scute (see Williams, I953, Plate 4b) , 
evident on both nuchal bones in which the sulci can be seen. Only in 
Deirochelys and the Pliocene C. wilUamsi among Florida emydids is the 
first vertebral scute regularly this wide. This condition occurs as an 
uncommon variant in both C. nelsoni (e.g., DRJ 225) and C. floridana 
(e.g., DRJ 255). However, the nuchal scute underlap is much longer than 
that of C. WilUamsi. In conclusion, the nuchal bones represent either 
Deirochelys or an undescribed species of Chrysemys that combines features 
of two Pliocene species, C. oaelata and C. wilUamsi. Interestingly, 
this combination of characters is common today in the Central American 
turtle, C. ornata. 



109 



Discuss ion 

Williams (1953) tentatively assigned the Thomas Farm Chvysemys to the 
floridana group on the basis of shell rugosity and other unspecified 
details. Rose and Weaver (I966) rightly pointed out that the pleural 
bone rugosity bore stronger resemblance to that of C. nelsoni and C. 
caelata and, based on this and two other characters that must be questioned 
on the grounds that they may have been drawn from Deirohelys, referred the 
Thomas Farm fossils to the nelsoni-whviventvis complex. 

It is not yet clear which of these two viewpoints is correct. 
Indeed, there is no evidence to indicate that these two lines within the 
subgenus Pseudemys (i.e., nelsoni-vuhviventvis and flovidnna-concinna) 
had diverged by the Miocene. (For reasons given in Chapter IV, I cannot 
agree with Rose and Weaver's [I966] hypothesis of a close relationship 
between C. nelsoni and C. soripta.) in fact, an Oligocene or early 
Miocene turtle like that at Thomas Farm, with strong resemblance to C. 
ovnata, would seem sufficiently generalized to give rise not only to both 
lines of Pseudemys {rubviventris and floridana) but to Trachemys as well. 
A final pal eoecolog lea 1 note may be added. Among freshwater emydid 
turtles, a rugose carapace and relatively long nuchal scute underlap seem 
to be associated with occupation of quiet waters, and a smooth carapace 
and short nuchal scute underlap with more strongly flowing water (these 
are only two of what appears to be a complex of habitat-related 
morphological characters). Regardless of generic affiliations, the nuchal 
scute underlap and carapacial rugosity of the Thomas Farm emydid fossils 
therefore imply that the Lower Miocene environment of Thomas Farm included 
slow-moving or still water, as previously suggested by Auffenberg (I963). 



Figure 23. Dorsal and visceral surfaces of second and third 
neural bones (UF 21933) of Chrysemys from Thomas 
Farm (x 2.2). 




Ill 




APPENDIX 
FOSSIL LOCALITIES CONTAINING DEIEOCHELYS 



Miocene: Arikareean site 
Thomas Farm, Gilchrist County. 

Auffenberg (1963a) reviews the geology and literature pertinent to 
this site in addition to discussing its ophidian fauna. He interprets 
the site as representing a lower Miocene fissure fill and points out 
biotic evidence for the presence of slow-moving or still water during 
that time. 

Pliocene: Hemphillian sites 
Haile Vl, Alachua County. 

One of a series of Pliocene sites assigned to the "Alachua Formation" 
of Florida, parts of its pa leoherpetofauna have been treated by Auffenberg 
(1955, 1963a), Coin and Auffenberg (1955), and D. Jackson (Chapter III). 
Auffenberg (1963a) discusses the stratigraphy of the deposit and states 
that it represents an ancient stream bed. The site lies approximately 
26 m above present sea level. 

Love Bone Bed, Alachua County. 

This previously unreported site (29°33'N, 82°31'W; Sec. 9, TllS 
R18E) near Archer, Alachua County, Florida, is named for Ronald Love who 
discovered it in 1974; it is now being excavated by the Florida State 
Museum under the supervision of S. David Webb. Preliminary strat igraphic 
studies reveal that the deposit represents the "Alachua clays" which were 

112 



113 



laid down in an ancient stream bed cut into uplifted Eocene Ocaia 
1 imestone. 

The presence of the horses Hipparion pUaatile Leidy and Nannippus 
ingenuus (Leidy). an early Osteoborus dog, the artiodactyls Synthetooeras 
and an advanced Cranio cer as , and an early saber-cat of the genus 
Bopbourofelis (S. D. Webb, pers. comm.), as well as the turtle Chrysemys 
caelata Hay (Chapter III), indicates an early Hemphill ian fauna roughly 
equivalent to that of McGehee Farm and Mixson's Bone Bed. 

McGehee Farm, Alachua County. 

An early Hemphill ian site in the Alachua Formation (Rose and Weaver, 
1966; Hirschfeld and Webb, I968), McGehee Farm is the type locality of the 
Pliocene emydine turtle Chrysemys wilUamsi (Rose and Weaver, I966), the 
tortoise Geoahelone alleni (Auffenberg, I966), and the chelydrid 
Maaroalemys auffenbergi (Dobie, I968), and has additionally yielded 
abundant material representing Chrysemys oaelata (Chapter III) and 
Trionyx sp. 

Mixson's Bone Bed, Levy County. 

The first known Pliocene deposit (Dall and Harris, 1892,- Leidy and 
Lucas, 1896) within the type section of the Alachua Formation (Simpson, 
1929a) of Florida, Mixson's Bone Bed is the type locality of Hay's (I908) 
Chrysemys oaelata (Chapter III). 

Pliocene; BJancan site 
Haile XVA, Alachua County. 

This deposit represents a former sinkhole filled with alternating 
coarse sands and clays. The fauna, assigned to the Aftonian interglac ial, 



11^ 



is characterized by a rich assortment of aquatic and terrestrial vertebrates, 
including the turtle Chvysemys platymarginata (Weaver and Robertson, I967) 
for which the site is the type locality. Although now Ik m above present 
sea level the presence of marine vertebrates in the fauna indicates higher 
sea level during deposition and a decided estuarine influence (Kinsey, 
197^; S. Webb, 197^). Robertson (1976) presents a detailed account of 
the stratigraphy and the mammalian fauna of the deposit. 

Pleistocene: Irvingtonian site 
Haile XVI, Alachua County. 

An undescribed pit (29°^0'40"N, 82°3^'20"W; Sec. 25, T9S, R17E) in 
the Haile limestone quarries (Ligon, I965) excavated by the Florida State 
Museum under the supervision of S. David Webb in May, 1973. The fauna 
appears to be i nterg lacial , of Irvingtonian age, and may represent the 
first known Yarmouthian deposit in Florida. A more detailed discussion 
of the deposit will accompany reports of faunal studies presently being 
conducted . 

Pleistocene: Rancholabrean sites 
Bradenton 51st Street, Manatee County. 

A coastal marsh 3 m above present sea level (S. Webb, 197^), the 
site is discussed by Simpson (1930a, b) and Auffenberg (1958; 1963a) and 
is now known to represent the Sangamonian interglacial (S. Webb, 197^). 

Coleman IMC, Sumter County. 

An undescribed deposit (Sec. 7, T20S, R23E) of Rancholabrean age 
(S. D. Webb, pers. comm. ) , this site, like previously reported Coleman 





115 


deposits (Martin 


197^), represents a filled sinkhole in the late Eocene 


Ocala Limestone; 


its surface lies approximately 24 m above present sea 


level . 





Kendrick lA, Marion County. 

A sinkhole-fissure fill near Kendrick, the deposit lies approximately 
2k m above present sea level. Kurten (I965) and Brodkorb (I959) assign 

the fauna to the illinoian or early Sangamonian although Auffenberg (I958) 

and S. Webb (197^) believe it to represent the Wisconsinan. 

Reddick IIC, Marion County. 

Approximately 2k m above present sea level, this inland deposit 
represents a Pleistocene sinkhole or fissure fill containing a Rancho- 
labrean fauna (S. D. Webb, 197^). 

St. Petersburg, Catalina Gardens, Pinellas County. 

A small, previously-unreported deposit (Sec. 12, T32S, R16E) at 
approximately present sea level, its fauna is apparently Rancholabrean 
in age (S. D. Webb, pers. comm.). 

Seminole Field, Pinellas County. 

A Pleistocene coasta] marsh three meters above present sea level, 
Simpson (1929a), Auffenberg (1958) and Kurten (I965) assign this site 
to the Wisconsinan glacial period. Cooke (1926) and Simpson (1929a) 
give accounts of the stratigraphy of the deposit. Simpson (1929a, b, 
1930b) lists the mammalian fauna, and Gilmore (1938), Brattstrom (1953) 
and Auffenberg (1963a) the snake fauna. 



116 



Vero, Indian River County. 

Only three meters above present sea level, this site is considered 
by most recent authors (Weigel, 1962; Auffenberg, 1963a; Webb. 197^) to 
represent the late Wiscons inan glacial period. Weigel (I962) discusses 
Its stratigraphy and vertebrate fauna (but does not include Deirochelys) 
and reviews the extensive literature pertaining to the site. 

Waccasassa River I. V. and VI. Levy County. 

Although S. Webb (1974) includes site VI in his chronology of 
Florida Pleistocene localities, neither of the other deposits which 
have yielded Deirochelys (l, Sec. 20 and V. Sec. 32. TI3S. R16E) has 
been mentioned. The sites are 6 m to 7 m above present sea level and 
contain Rancholabrean faunas (S. D. Webb, pers. comm.). 

Wall Company Pit, Alachua County. 

The stratigraphy of this small fissure deposit is briefly discussed 
by Auffenberg (1963a). Auffenberg (1958) tentatively assigns the deposit 
to a time between the illinoian glacial maximum and the Sangamonian 
interglacial maximum. 

Subrecent sites 
Nichol's Hammock. Dade County. 

Hirschfeld (I968) describes the geology, pal eoecology, and 
vertebrate fauna of this solution hole site. Her herpteofaunal list 
includes Deirochelys retioularia. 

Warm Mineral Springs, Sarasota County. 

Ferguson et al. (19^7) describe the hydrology and topography of 
this spring (Sec. 25, T39S. R20E. less than six meters above present sea 



17 



level) under the name of Warm Salt Spring. Clausen et al. (1975) 
present a detailed description of the geology of the deposit for which 
they report a radiocarbon age of approximately 10,000 years. Fossils 
are currently being excavated by the Florida Department of State under 
the direction of W. A. Cockrel 1 . 



LITERATURE CITED 

Adler, K. I968. Synonymy of the Pliocene turtles Pseudemys hilli Cope 
and Chrysemys limnodytes Galbreath. J. Herpetol . 1: 32-38. 

Agassiz, L. 1857. Contributions to the natural history of the United 
States of America. Vol. 1, 2. Little, Brown and Co., Boston. 
640 p. 

Alt, D. 1957. Pattern of post-Miocene eustatic fluctuation of sea level 
Geol. Soc. America, Southeast Sec. Program, p. I5. 

_, and H. K. Brooks. 196^. Age of the Florida marine terraces. 



J. Geol. 73: 406-4ll. 

Auffenberg, W. 1955. Glass lizards {Ophisaurus) in the Pleistocene and 
Pliocene of Florida. Herpetolog ica 11: I33-I36. 

• 1958. Fossil turtles of the genus Terrapene in Florida. Bull 

Florida State Mus. 3: 53-92. 

• 1963a. The fossil snakes of Florida. Tulane Stud. Zool 

10: I3I-2I6. 

• 1963b. Fossil testudinine turtles of Florida. Genera 

Geochelone and Flovidemys . Bull. Florida State Mus. 7: 53-97. 

• '966. A new species of Pliocene tortoise, genus Geochelone, 

from Florida. J. Paleontol. 40: 877-882. 



197^. Checklist of fossil land tortoises (Testud in idae) 



Bull. Florida State Mus. 18: 121-251. 

Baur, G. I889. The relationship of the genus Deiroohelys. Amer. Nat 
23: 1099-1100. 

Beyer, G. E. I9OO. Louisiana herpetology. Proc. Louisiana Soc. Nat. 
1897-1899: 25-46. 

Bickham, J. W. 1975- A cytosystemat i c study of turtles in the genera 
Clemmys, Mauremys and Sacalia. Herpetologica 31: 198-204. 

, and R. J. Baker. 1976. Chromosome homology and evolution of 



emydid turtles. Chromosoma (Berl.) 54: 201-219. 

Bjork, P. R. 1970. The Carnivora of the Hagerman local fauna (Late 
Pliocene) of southwestern Idaho. Trans. American Philosophical 
Soc. 60: 1-54. 

118 



19 



Boulenger. G. A I889. Catalogue of the chelonians, rhynchocephal ians 
and crocodiles in the British Museum (Natural History). Brit sh 
Mus. (Nat. Hist.), London. 311 p. 

Bramble DM 1973. Media dependent feeding in turtles. Amer. Zool. 



13: 13^2 



I97J. Emydid shell kinesis: Biomechanics and evolution. 



Copeia 197/*: 707-727 

Brattstrom B. H 1953. Records of Pleistocene reptiles and amphibians 
from Florida. Quart. J. Florida Acad. Sci. 16: 243-2^8. 

Brodkorb. P. 1958. Fossil birds from Idaho. Wilson Bull. 70: lH-lkl. 

1959. The Pleistocene avifauna of Arredondo, Florida. Bull. 



Florida State Mus. k: 269-291. 

Brooks H. K I966. Geological history of the Suwannee River. In N K 
Olson (ed.) Geology of the Miocene and Pliocene series in the ' 
north Florida-south Georgia area. Atlantic Coastal Plain Geol 
Assoc. 7th Ann. Field Conf. and Southeastern Geol. Soc 12th 
Ann. Field Conf. 

Cagle F. R. I952. The status of the turtles Gvaptemys pulchra Baur and 
Gvaptemys barbour^ Carr and Marchand, with notes on their natu al 
history. Copeia 1952: 223-23^. 

.. 1953. Two new subspecies of Graptemys pseudogeographica. 



Occas. Pap. Mus. Zool., Univ. Michigan, no. 5^6: I-I7. 

Z^ol ^?^^i67-T86"^'' species of the genus Graptemys. Tulane Stud. 

Campbell. H^W.^^,96_9^^^The unsung chicken turtle. int. Turtle. Tortoise 

^^'''' m'v^' JP" "^"^'^°°'< °^ turtles. Comstock Publ . Assoc, Ithaca 
N . r . i)42 p. ' 

., and L. J. Marchand. 19^2. A new turtle from the Chipola 



River, Florida. Proc. New England Zool. Club 20: 95-100. 
^^^"^^'h-eek'^' ^^^°' ^PP^-" f'"^""^ f'-ogs from Idaho. Copeia I97O: 

Clausen C. J. H. K Brooks, and A. B. Wesolowsky. 1975- Florida spring 
confirmed as 10,000 year old early man site. The Florida 
Anthropologist 28, no. 3, part 2: I-38. 

Conant, R. 1975. A field guide to reptiles and amphibians of eastern 
and central North America. Houghton Mifflin Co., Boston. 429 p. 



120 



Cooke, C. W. 1926. Fossil man and Pleistocene vertebrates in Florida 
Amer. J. Sci., 5th Ser. 12: ^41-452. 

1945. Geology of Florida. Florida Geol. Surv. Bull. 29: 



1-339. 

Cope, E. D. 1878. Descriptions of new extinct Vertebrata from the Upper 
Tertiary and Dakota formations. Bull. U.S. Geoqr. Surv. Territories 
h: 379-396. 'Lor.es, 

Crenshaw, J. W. I955. The ecological geography of the Pseudewys floridana 
complex in the southeastern United States. Ph.D. dissertation, Univ. 
Florida, Gainesville. 205 p. 



1965. Serum protein variation in an interspecies hybrid 



of turtles of the genus Pseudemys. Evolution 19: I-I5. 

Dall, W. H., and G. D. Harris. I892. Correlation papers: Neocene. 
Bull. U.S. Geol. Surv. no. 84, 3^9 p. 

Dobie, J. L. 1968. A new turtle species of the genus Macvoalemye 

(Chelydridae) from the Florida Pliocene. Tulane Stud. Zool Bot 
15: 59-63. 

1972. Correction of distributional records for Graptemys 



swarm 



barbouri and Graptemys pulchra. Herpetol. Rev. k: 23 

Donn, W. L. , W. R. Farrand, and M. Ewing. I962. Pleistocene ice volumes 
and sea-level lowering. J. Geol. 70: 206-214. 

Ernst, C. H., and R. W. Barbour. 1972. Turtles of the United States. 
Univ. Press of Kentucky. Lexington. 347 p. 

Feduccia, J. A. 1967. Ciconia maltha and Gvus amerlcana from the Upper 
Pliocene of Idaho. Wilson Bull. 79: 3I6-3I8. 

Ferguson, G. E., C. W. Lingham, S. K. Love, and R. 0. Vernon. 1947. 
Springs of Florida. Florida Geol. Surv. Bull. no. 31. 1 96 p. 

Fichter, L. S. 1969. Geographical distribution and osteological variation 
In fcssils and Recent specimens of two species of Kinostemon 
(Testudines) . J. Herpetol. 3: II3-II9. 

Folkerts, G. W. , and R. H. Mount. I969. A new subspecies of the turtle 
Graptemys nigrlnoda Ca'gle. Copeia I969: 677-682. 

Frey, D. G. I965. Other i nvertebrates--An essay in biogeography , p. 

613-631. In H. E. Wright, Jr. and D. G. Frey (eds.) The Quaternary 
of the United States. Princeton Univ. Press, New Jersey. 

Gaffney, E. S. I972. An illustrated glossary of turtle skull nomenclature. 
Amer. Mus. Novitates No. 2486: I-33. 



121 



Galbreath, E. C. 19^*8. A new extinct emydid turtle from the Lower 
Pliocene of Oklahoma. Univ. Kansas Publ. Mus. Nat. Hist. 1: 
265-280. 

Gibbons, J. W. 1969- Ecology and population dynamics of the chicken 
turtle, Deivoahelys retioularia . Cope i a 1969: 669-676. 

1970. Terrestrial activity and the population dynamics of 



aquatic turtles. Amer. Midi. Nat. 83: kOk-k]k. 

Gilmore, C. W. 1933- A new species of extinct turtle from the Upper 
Pliocene of Idaho. Proc. U.S. Nat. Mus. 82: I-7. 

1938. Fossil snakes of North America. Geol. Soc. Amer., 



Spec. Pap. 9: 1-96. 

Goin, C. J., and W. Auffenberg. 1955. The fossil salamanders of the 
family Sirenidae. Bull. Mus. Comp. Zool . 113: ^97-514. 

Gray, J. E. I87O. Supplement to the catalogue of shield reptiles in the 
collection of the British Museum. Part I. Testudinata (tortoises). 
British Mus., London. 120 p. 

Harper, F. 19^0. Some works of Bartram, Daubin, Latreille, and 

Sonnini, and their bearing upon North American herpetolog leal 
nomenclature. Amer. Midi. Nat. 23: 692-723. 

Hay, 0. P. 1902. Bibliography and catalogue of the fossil vertebrata of 
North America. Bull. U.S. Geol. Surv. , no. 179. 868 p. 

1908. Fossil turtles of North America. Carnegie Inst. 



Washington Publ. 75, 568 p. 

• 1916. Descriptions of some Floridian fossil vertebrates, 

belonging mostly to the Pleistocene. Ann. Rept. Florida Geol, 
Survey. 8: 39-76. 

• 1923. The Pleistocene of North America and its vertebrated 

animals from the states east of the Mississippi River and from the 
Canadian Provinces east of longitude 95°. Carnegie Inst. Washington 
Publ. 322. i*99 p. 

1927. The Pleistocene of the western region of North America 



and its vertebrates animals. Carnegie Inst. Washington Publ. 322b. 
3^*6 p. 

Hellier, T. R. , Jr. I967. The fishes of the Santa Fe River system. 
Bull. Florida State Mus. Biol. Sci. 2: 1-46. 

Hibbard, C. W. , D. E. Ray, D. E. Savage, D. W. Taylor, and J. E. Guilday. 
1965. Quarternary mammals of North America, p. 509-525. In H. E. 
Wright, Jr. and D. G. Frey (eds.) The Quarternary of the United 
States. Princeton Univ. Press, New Jersey. 



122 



Hirschfeld, S. E. I968. Vertebrate fauna of Nichol's Hammock, a natural 
trap. Quart. J. Florida Acad. Sci. 31: I77-I89. 

, and S.D. Webb. I968. PI io-Plei stocene megalonychid sloths of 



North America. Bull. Florida State Mus. 12: 213-296 

Hoffman, C. K. I89O. Reptilien I, Schi Idkroten. 6(3): 1-442, 48 pi. In 
Klassen und Ordnungen des Thei r-Reichs . H. G. Bronn, ed. Leipzig. 

Holman, J. A. I967. A Pleistocene herpetofauna from Ladds, Georgia 
Bull. Georgia Acad. Sci. 25: 154-166. 

. 1968. Upper Pliocene snakes from Idaho. Copeia I968: I52-I58. 

Jackson, C. G. I959. The osteology of the chicken turtle, Deirochelys 
vetvculavva (Latreille), with reference to the genera, Chrysemys 
and Emydo%dea. Unpubl. M.S. thesis, Univ. Florida, Gainesville. 
101 p. 

. 1964. The status of Deirochelys floridana Hay with comm.ents 

on the fossil history of the genus. Tulane Stud. Geol. 2: IO3-IO6. 

. 1974a. The status of Trachemys jarmani Hay with clarification 

of the fossil record of Deirochelys. Copeia 1974: 536-537. 

. 197'*b. An unusual pattern of cervical central articulation in 

Deirochelys reticularia. Copeia 1974: 788. 

' ^^'^ J- M. Kaye. I974. The occurrence of Blanding's turtle, 

Emydoidea blandingii, in the late Pleistocene of Mississippi 
Herpetologica 30: 417-419. 

Kinsey, P. E. 1974. A new species of Mylohyus peccary from the Florida 
Early Pleistocene, p. I58-I69. In S. D. Webb (ed.) Pleistocene 
mammals of Florida. Univ. Florida Press, Gainesville. 

Kurten, B. I965. The Pleistocene Fel idae of Florida. Bull. Florida 
State Mus. 9: 215-273. 

Leidy, J., and F. A. Lucas. I896. Fossil vertebrates from the Alachua 
clays of Florida. Trans. Wagner Free Inst. Science 4: I-6I. 

Ligon, J. D. I965. A Pleistocene avifauna from Haile, Florida. Bull 
Florida State Mus. 10: I27-I58, 

Loveridge, A., and E. E. Williams, 1957. Revision of the African 

tortoises and turtles of the suborder Cryptodira. Bull. Mus Como 
Zool. 115: I63-557. 

McCoy, C J. 1973. Emydoidea; E. blandingii. Cat. Amer. Amphib. Rept 
136.1-136.4. 



123 



McDowell, S. B. 196^*. Partition of the genus Clemmys and related 

problems in the taxonomy of the aquatic Testud in idae. Proc Zool 
Soc. London 1^3: 239-279. 

McKown, R. R. I972. Phylogenetic relationships within the turtle genera 
Graptemys and Malaclemys. Ph.D. dissertation. Univ. Texas 
Austin. Ill p. ' 

Martin, R. A. I974. Fossil mammals from the Coleman MA fauna, Sumter 

County, p. 35-99. In S. D. Webb (ed.) Pleistocene mammals of Florida. 
Univ. Florida Press, Gainesville. 

Martof, B. S. I963. Some observations of the herpetofauna of Sapelo 
island, Georgia. Herpetolog ica 19: 70-72. 

Mayr, E. I969. Principles of systematic zoology. McGraw-Hill Book Co. 
New York. 428 p. 

Merkle, D. A. I975. A taxonomic analysis of the Clemmys complex (Reptilia: 
Testudines) utilizing starch-gel electrophoresis. Herpetoloqica 
31: 162-166. 

Miller, R. R. I965. Quaternary freshwater fishes of North America, p. 

569-581. In H. E. Wright, Jr., and D. G. Frey (eds.) The Quaternary 
of the United States. Princeton Univ. Press, New Jersey. 

and G. R. Smith. 1967. New fossil fishes from PI io-Plei stocene 



Lake Idaho. Occas. Pap. Mus. Zool. Univ. Michigan (65^): 1-24. 

Milstead, W. W. ^ 1956. Fossil turtles of Freisenhahn Cave, Texas, with 

the description of a new species of Testudo. Copeia 1956: 162-171. 

. 1967. Fossil box turtles {lervapene) from central North America 

and box turtles of eastern Mexico. Copeia I967: 168-179. 

. 1969. Studies on the evolution of box turtles (genus Tervapene) 

Bull. Florida State Mus. 14: 1-113. 

' 3nd D. W. Tinkle. I967. Tervapene of western Mexico, with 

comments on the species groups in the genus. Copeia I967: I8O-I87. 

Moll, E. 0., and J. M. Legler. 1971. The life history of a Neotropical 
slider turtle Pseudemys scripta (Schoepff) in Panama. Bull. Los 
Angeles Co. Museum Nat. Hist., Science, no. 11: 1-102. 

Mount, R. H. 1972. Distribution of the worm snake Carphophis amoenus 
(Say) in Alabama. Herpetolog ica 28: 263-266. 

1975. The reptiles and amphibians of Alabama. Auburn Univ. 



Agric. Exp. Sta. , Auburn, Alabama. 347 p. 

' and G. W. Folkerts. I968. Distribution of some Alabama 

reptiles and amphibians. Herpetolog ica 24: 259-262. 



]2k 



Murray, B. G. , Jr. I967. Grebes from the Late Pliocene of North 
America. Condor 69: 277-288. 

Neill, W. T. ig'+S. Hibernation of amphibians and reptiles in Richmond 
County, Georgia. Herpetolog i ca h: 107-114. 

• 1958. The occurrence of amphibians and reptiles in salt- 
water areas, and a bibliography. Bull. Marine Sci. Gulf Caribbean 
8:1-97. 

Olsen, S. J. 1964a. The st rat igraphi c importance of a Lower Miocene 
vertebrate fauna from North Florida. J. Paleontol. 38: 477-482. 

1964b. Vertebrate correlations and Miocene stratigraphy of 



North Florida fossil localities. J. Paleontol. 38: 600-604 

Parsons, T. S. I96O. The structure of the choanae of the Emydinae 

(Testudines, Testud in idae) . Bull. Mus. Comp. Zool . 123: II3-I27. 

Patton, T. H. I969. An Oligocene land vertebrate fauna from Florida. 
J. Paleontol. 43: 543-546. 

Pope, C. H. 1939. Turtles of the United States and Canada. Alfred A. 
Knopf, New York. 343 p. 

Preston, R. E. I966. Turtles of the Gilliland faunule from the 

Pleistocene of Knox County, Texas, Pap. Michigan Acad. Sci., Arts 
and Letters. 51 : 221-239. 

971. Pleistocene turtles from the Arkalon local fauna of 



southwestern Kansas. J. Herpetol . 5: 208-211. 
_, and C. J. McCoy. 1971. The status of Emys twentei Taylor 



(Reptilia: Testud in idae) based on new fossil records from Kansas 
and Oklahoma. J. Herpetol. 5: 23-3O. 

Pritchard, P. C. H. I967. Living turtles of the world. T. F. H. Publ . , 
Jersey City, New Jersey. 288 p. 

Puri, H. S., and R. 0. Vernon. 1964. Summary of the geology of Florida 
and a guidebook to the classic exposures. Florida Geol . Surv 
Spec. Publ. 5: I-3I2. 

Robertson, J. S. 1976. Latest Pliocene mammals from Haile XVA, Alachua 
County, Florida. Bull. Florida State Mus., Biol. Sci. 20: 111-186. 

Rogers, K. L. 1976. Herpetofauna of the Beck Ranch local fauna (Upper 
Pliocene: Blancan) of Texas. Publ. Mus., Michigan State Univ., 
Paleontological Ser. 1: 167-200. 

Rose, F. L., and W. G. Weaver, Jr. I966. Two new species of Chrysemys 
(= Pseudemys) from the Florida Pliocene. Tulane Stud. Geol 5- 
41-48. 



125 



Schwartz, A. 1956. Geographic variation in the chicken turtle 

Deiroohelys reticularia Latreille. Field fana (Zool.) Ik: ^61-503. 

Shah, R. v. 1963- The neck musculature of a cryptodire (Deirockelys) 
and a pleurodire [Chelodina) compared. Bull. Mus. Comp. Zool. 
129: 3^3-368. 

Simpson, G. G. 1929a. The extinct land mammals of Florida. Ann. Rept. 
Florida Geol . Surv. 20: 229-279. 

• 1929b. Pleistocene mammalian fauna of the Seminole Field, 

Pinellas County, Florida. Bull. Amer. Mus. Nat. Hist. 56: 561-599. 

1930a. Tertiary land mammals of Florida. Bull. Amer. Mus. 



Nat. Hist. 59: 1^*9-211 

• 1930b. Additions to the Pleistocene of Florida. Amer. Mus. 

Nov. No. '*06: 1-14. 

Smith, H. M. , R. B. Smith, and H. L. Sawin. 1976. The generic name of 
the Neotropical semi terrestrial emydine turtles. Trans. Kansas 
Acad. Sci, 77: 211-217. 

Swift, C. C. 1970. A review of the eastern North American cyprinid 

fishes of the Notropis texanus species group (subgenus Alburnops) , 
with a definition of the subgenus Hydrophlox, and materials for a 
revision of the subgenus Alburnops. Ph.D. dissertation. Florida 
State Univ., Tallahassee. 515 p. 

Taylor, D. W. I966. Summary of North American Blancan nonmarine 
mollusks. Malacologia h: 1-172. 

Taylor, E. H. 19^3. An extinct turtle of the genus Emys from the 
Pleistocene of Kansas. Univ. Kansas Sci. Bull. 29: 2^9-254. 

Tinkle, D. W. 1962. Variation in shell morphology of North American 

turtles. I. The carapacial seam arrangements. Tulane Stud. Zool. 
9: 331-349. 

Vernon, R. 0. 1951- Geology of Citrus and Levy Counties, Florida. 
Florida Geol. Survey Bull. 33: I-256. 

Waagen, G. N. 1972. Musk glands in Recent turtles. Unpubl . M.S. 
thesis, Univ. Utah, Salt Lake City. 65 p. 

Weaver, W. G. , Jr., and J. S. Robertson. I967. A re-evaluation of fossil 
turtles of the Chrysemys soripta group. Tulane Stud. Geol. 5: 
53-66. 

_, and F. L. Rose. I967. Systematics, fossil history, and evolution 



of the genus Chrysemys . Tulane Stud. Zool. I4: 63-73. 



126 



Webb. G. J. W and C. R. Johnson. 1972. Head-body temperature differences 
in turtles. Comp. Biochem. Physiol. 43A: 593-611. 

Webb, R. G. 1950. Range extension of the chicken turtle in Oklahoma. 
Herpetologica 6: 137-138. 

Webb. S. D 1964. The Alachua Formation, p. 22-29- In Guidebook 1964 
Field Trip Soc. Vert. Paleontology in Central Florida. Univ 
Florida, Gainesville. 

1969. The Pliocene Canidaeof Florida. Bull. Florida State 



Mus. U: 273-308 

. 1974. Chronology of Pleistocene land mammals in Florida, p. 

5-31. In S. D. Webb (ed.) Pleistocene mammals of Florida. Univ 
Florida Press, Gainesville. 

, and N. Tessman. I968. A Pliocene vertebrate fauna from low 

elevation in Manatee County, Florida. Amer. J. Science 266: 777-811 

Weber. W. A. I965. Plant geography in the southern Rocky Mountains 
p. 453-468. In H. E. Wright, Jr., and D. G. Frey (eds.) The ' 
Quaternary of the United States. Princeton Univ. Press. New Jersey. 

Weigel R. D. 1962. Fossil vertebrates of Vero, Florida. Florida Geol 
Surv. Spec. Publ . 10: 1-59. 

Wharton. C. H. , T. French, and C. Ruckdeschel. 1 973. Recent range- 
extensions for Georgia amphibians and reptiles. HISS News-J. 

White, T. E. 1929. The osteology of the recent turtles of central 

North America. Unpubl . M.A. thesis, Univ. Kansas, Lawrence. 307 p. 

White, W. A. 1970. The geomorphology of the Florida peninsula. Florida 
Bureau of Geol., Geol. Bull. 51. 

Williams, E. E. 1950. Variation and selection in cervical central 
articulations of living turtles. Bull. Amer. Mus. Nat Hist 
94: 505-562. 

. 1953. A new fossil tortoise from the Thomas Farm Miocene of 



Florida. Bull. Mus. Comp. Zool. 107: 537-554. 

. 1956. Pseudemys saripta aallirostris from Venezuela with a 

general survey of the saripta series. Bull. Mus. Comp. Zool 
115(5): 145-160. 

Wilson, J. W. . III. 1975. Morphological change as a reflection of 
adaptive zone. Amer. Zool. 15: 363-370. 



127 



Wood, R. C, and B. Patterson. 1973. A fossil trionychid turtle from 
South America. Breviora Mus. Comp. Zool. No. ^05: 1-10. 

Yerger, R. W. , and K. Relyea. I968. The flat-headed bullheads (Pisces: 
Ictaluridae) of the southeastern United States, and a new series of 
Ictaliirus from the Gulf Coast. Copeia I968: 361-38^. 

Zakrzewski, R. J. I969. The rodents from the Hagerman local fauna. 
Upper Pliocene of Idaho. Contrib. Mus. Paleontology, Univ. 
Michigan. 23: I-36. 

Zug, G. R. 1966. The penial morphology and the relationships of 

cryptodiran turtles. Occ. Paper Mus. Zool., Univ. Michigan No 
6^7: 1-24. 

1969. Fossil chelonians, Chrysemys and Clernmys, from the 



Upper Pliocene of Idaho. Great Basin Naturalist 29: 82-87 

» ^nd A. Schwartz. 1971. Deiroohelys: D. retioulavia. Cat. 

Amer. Amphib. Rept. 107. 1-10?. 3. 



BIOGRAPHICAL SKETCH 

Dale Robert Jackson was born in Boston, Massachusetts, on December 
17, 19^9. He entered Eastern Illinois University in September, I967, 
and received the degree of Bachelor of Science with a major in zoology 
in June, 1971. Setting out to seek his fortune, the naive midwestern 
youth journeyed southward and, in September that same year, entered the 
Graduate School of the University of Florida. There he pursued the 
degree of Doctor of Philosophy in the field of zoology. 

He is a member of the American Society of Ichthyologists and 
Herpetologists, American Society of Naturalists, American Society of 
Zoologists, American Association for the Advancement of Science, and 
the National Audubon Society. 

He is unmarried, has no children to his knowledge, and is now 
seeking happiness rather than fortune. 



128 



I certify that I have read this study and that in my opinion it 

conforms to acceptable standards of scholarly presentation and is fully 

adequate, in scope and quality, as a dissertation for the degree of 
Doctor of Philosophy. 



Walter Auffenber^^, Chi'j/man 
Professor of iQjsAoqy '''^ 




I certify that I have read this study and that in my opinion it 

conforms to acceptable standards of scholarly presentation and is fully 

adequate, in scope and quality, as a dissertation for the degree of 
Doctor of Philosophy. 



tan 



^l/lA/^^ 



Carmine A. Lancianl 

Associate Professor of Zoology 



I certify that i have read this study and that in my opinion it 
conforms to acceptable standards of scholarly presentation and is fully 
adequate, in scope and quality, as a dissertation for the degree of 
Doctor of Philosophy. 




Frank G. Nordl ie 
Professor of Zoology 



I certify that I have read this study and that in my opinion it 
conforms to acceptable standards of scholarly presentation and is fully 
adequate, in scope and quality, as a dissertation for the degree of 
Doctor of Philosophy. 




Assistant Professor of Geology 



I certify that I have read this study and that in my opinion it 

conforms to acceptable standards of scholarly presentation and is fully 

adequate, in scope and quality, as a dissertation for the degree p^- 
Doctor of Philosophy. 




S. David'Welib " \^ V 
Professor of Zoology 



This dissertation was submitted to the Graduate Faculty of the Department 
of Zoology in the College of Arts and Sciences and to the Graduate Council, 
and was accepted as partial fulfillment of the requirements for the degree' 
of Doctor of Philosophy. 

August, 1977 



Dean, Graduate School 



UNIVERSITY OF FLORIDA 



3 1262 08553 2884